Probability System Theory and Money Management | Forex Factory

Options

Go to
Permalinks
Print Thread

Similar Threads

Alternative Trade Management/Money Management Methods 3 replies

Money Management Theory 94 replies

Money Management / Risk Management 24 replies

Probability, Pyramiding and Money Management 37 replies

Lessons in directional probability and risk management 29 replies

Probability System Theory and Money Management

  • #1
  • First Post: Dec 8, 2006 11:37pm Dec 8, 2006 11:37pm
I feel I need to create this thread because the ideas it discusses are beginning to get off topic of the thread in which they were born. The original thread is located here

This thread is to discuss using mathematical probabilities to your advantage in a successful system. I'll try to make it simple to explain.

If you have a system that utilizes a 100 TP and 100 SL it should have a 50/50 chance of winning if trades are entered at random over time. Correct? It'd be like flipping a coin. The TP and SL would have to be adjusted since you are closer to the SL as soon as you enter a trade. But let's keep it simple theory for now.

If we enter a trade that is equidistant from the SL and TP then there should be a 50% chance of being correct. I will then assume that by actually looking at the general trend or market activities of a currency that we can give us a system that is more than 50% successful or at least 50% successful if my first assumption was incorrect. Does anyone disagree so far?

The probabilities haven't come into play yet though. I'm asking now what are the chances of a 50% correct system having consecutive losers?

If I remember correctly from college we have a 50/50 chance on each trade to be right and wrong when we view them as independent events. When we try to find the probability of having 5 losing trades in a row, however, the chance of that happening is then:

.5 X .5 X .5 X .5 X .5 = .03125 or 3.125%. Doesn't happen too often, although still very possible. If our system is successful 60% of the time due to our simple chart reading predictions then the chance of having 5 losers would be 40%^5th power. .4 X .4 X .4 X .4 X .4 = .010124 or 1.0124%. Still possible, although with some discretion we've effectively lowered the risk of losing 5 times in a row by a factor of 3.

On the other thread, I was using a 100 SL and 50 TP to increase the success rate. The Risk:Reward was skewed to the losing side. I am starting this new thread with an equal risk vs reward to give us a 50/50 chance on each random trade. I wouldn't use a 1000 TP with 100 SL because the probability of getting a successful trade is very low and to use the following money management strategy would require too large a bank account to be reasonable.

Here's the idea. After a losing trade, double up on the next trade to cover the losses of the losing trade when you win this one. With a SL and TP that give us equal gains versus losses, doubling up on the winners allows us to cover the losses on the losers and still profit on the winning trade. So it would make it seem as if you had never taken the bad trade at all.

Here are the probabilities of getting x amount of losers in a row in a 50% system:

2- .25 or 25%
3-.125 or 12.5%
4-.0625 or 6.25%
5- .03125 or 3.125%
6- .015625 or 1.5625%
7-.0078125 or .78125%
8- .00390625 or .390625%
9- .001953125 or .1953125%
10 - A really small number

This means out of 1000 trades you can expect to lose 9 times in a row 1.9 times. If your system hasn't provided enough profit after 1000 trades to survive that then you're trading the wrong system.

Now let's see what doubling up would require. Our value is 1 for your typical lot size per trade. To double up on consecutive trades to get to the next you would need to risk

1 - 2 - 4 - 8 - 16 - 32 - 64 - 128 - 256 - 512.


This means if your typical trade size is 5 minilots then you need to trade:
5-10-20-40-80-160-320-640-1280...etc.

So you can see that it adds up quickly and it would be very unwise to trade 5 minilots on a typical trade unless you have very high leverage as well as a large bank account. I would suggest trying the idea on micro micro lots first.

I'm sure this idea of doubling down after losing trades has been discussed before, but I am not sure what it is generally referred to as. It's not martingale because it doesn't require you continue to fight with a losing trade. It's not pyramiding because you're not adding to trades when you're winning. It's more of the opposite. In fact I think it would be very unwise to trade with the pyramid strategy for winning trades, because as my numbers above show, the next consecutive winner, just like consecutive losers, is very unlikely and you're probably going to wipe away a lot of your profits when that loser hits.

There are two parts to the thread I think that should be discussed.

1. The doubling down strategy. With a risk:reward of 1:1 the system is very possible to use if you use very strict money management and understand that in the beginning money will be coming slowly, but it will be coming consistently. The strategy makes it seem as if there were no losing trades, remember? This is a law of large numbers system that requires you to be in the game for the long haul and not to make a quick buck to buy holiday gifts.

and

2. The strategy of the actual trading system. Equal profits and losses. Randomness. In the end how much does all of our research really increase our chances of success? The market only goes up or down. You can claim it goes sideways, but it really doesn't. Sideways movement still has pips up and down inside of it. So then how much validity is there in my assumptions that if we randomly place trades with equidistant TP and SL's that there is a 50/50 chance on each trade? And therefore, very limited possibilities of a random system having many consecutive losses.

I already have many ideas for how to trade this in my head. Yes, these ideas could be traded extremely easily by an EA. In fact since no indicators are required it could be an "always in" system traded very easily on a grid. I'd rather not start discussions about that yet.

Please just post about the topics already presented. Am I analyzing the market correctly? I'm sure plenty will bash me for presenting such ideas since doubling down is very risky and that the market is not necessarily random since technical analysis is a proven concept.

My response to those are that doubling down is risky. Does that mean someone with a $10,000 account can't institute this system when playing with pips valued at 1 cent and high leverage? I think they'd be able to afford to increase their lot size by 1028 times to make each pip worth $10.28. To get to 1028 their normal trading size would require 11 losing trades in a row. Imagine how unlikly that is.

And my response to the market not being random is the following. S/R is a proven concept. So are elliot waves. So are RSI crosses, channels, divergences, candles, head and shoulders, cup and handles, etc. etc.. The list goes on. The market is not random when we look at it in hindsight and know what tool to apply to see why it jumped 50 pips. Was it crossing the RSI, breaking through a previous resistance area, hitting a bottom channel? What tools the market follows are not constant and in that I claim the market is random. At least by it not following any one tool exactly then it is random and we can never tell where it will go. Why was this month a 15 year high on gbp/usd? How did it break through every previous resistance level in the past 14 years to get where its at now?

We're not in this business to predict and play the improbable. I doubt anyone here would have had the guts to enter a long trade in march of this year and hold out till these last two weeks. We would have packed up bags after bagging an easy 2000 pips several months ago and called it a year on gbp/usd. That is exactly what my idea for the system is about. I'm not here to predict every nook and cranny, retrace, peak, or valley. I'm trying to play the probable scenerios to bring in probable profits. Probability says a random, or even discretionary system is severely unlikely to lose 11 times in a row.

One thing I said on the original thread that rings true and cracked me up was:

Even if you were trying to be wrong you'd still probably win since the market wouldn't let you be 100% right about being wrong.

Think about all that and get back to me. Matt
  • #2
  • Dec 8, 2006 11:57pm Dec 8, 2006 11:57pm
To those that would be turned off by sacraficing your typical trade size to survive using these principles think about this.

What if you had enough leverage, access to micro lots, and enough of a bankroll to play the doubling down as if your SL was 100 and TP was 50?

To break even, and only break even on losing trades in that system your double down line would look like this:

1 - 2 - 6 - 18 - 54 - 162 - 486 - 1458 - etc.

But if you apply this to your equal losses gains system of 100 SL and 100 TP then not only does it make it seem like the losing trades never happened, it makes it seem as if the losing trades were actually winning trades. Not only that. Look at what happens if you lost 3 trades in a row and won on the fourth. You would have lost 900 pips, but gained 1800 for a total profit of +900 pips.

So on four trades you made 900 pips, where if you had won all 4 you would have only made 400. You were willing to risk more on each consecutive trade knowing statistically that it was very improbable to continue to lose. So you almost want the system to lose a couple times in a row once in a while. It increases your probability of success and more profit potential.

Like I said, money will be slow in the beginning, but eventually when you have enough to handle the risk you slowly increase the multiplication factor to bring in massive profits. Imagine making 5 bad trades in a row and winning on the 6th. You would have lost 8100 pips total, but banked 16200. Do you see why I insist this type of money management be played out on a micro micro lot broker? But hey, you just made $81 using 1 cent pips. Compound compound compound. Invest invest invest. Eventually the money will come.

Anyways, I need criticism. What am I missing? In the last thread I completely missed the concept that using a 100 SL and 50 TP would require more than just doubling on the trades to break even. This thread changes the variables to make it more plausible with a bankroll and lot sizes. HELP! Matt
  • #3
  • Dec 9, 2006 12:11am Dec 9, 2006 12:11am
Thanks ... still trying to digest them ...

.
  • #4
  • Dec 9, 2006 12:15am Dec 9, 2006 12:15am
I believe this is a martingale strategy, doubling up to get back loss plus profit. Whatever you want to call it here is one problem. Your math basically states that the chances of x happening are y, and y being a very small %. In your example you use 1000 coin flips or trades. The problem with this system is that those 9 times out of 1000 will wipe you out. You state that you must use good money management, however your money management has already been defined. - Trade it all the time and 1 to 1 risk to reward(although you did forget the zero and double zero known as the spread). Lastly how about the house limit. Can you trade 1032 lots, and if you lose can you trade 2064, and if you lose again, will you be able to trade 4128 lots. Howabout slippage with these large plays? Can you ever realistically move up in lot size. All of these factors make this an unfavorable game and the only way to win an unfavorable game is to bet big and stop playing when you win. If you plan on trading for a living, you will experience 9 losses in a row using this system, maybee more. This string of losses may not be today nor tomorrow, but it will come if you strictly follow this, and when it does come, you will get wiped out. Not trying to harp on your system, just some concerns.
Man who scratches ass should not bite fingernails
  • #5
  • Dec 9, 2006 12:50am Dec 9, 2006 12:50am
  •  tesla
  • Joined Oct 2006 | Status: Friendly Neighborhood Programmer | 533 Posts
What a coincidence! I was talking to someone just two days back about the same sort of thing.

Quoting permanentjaun
You were willing to risk more on each consecutive trade knowing statistically that it was very improbable to continue to lose. So you almost want the system to lose a couple times in a row once in a while. It increases your probability of success and more profit potential.
It's quite possible you're aware of this already, I'm just going to mention it because the wording you used above sets off warning bells in my head:

Each trade is a distinct entity. Losing on three trades in a row does not make you more likely to win on the fourth. That fourth trade will still have a 50/50 probability of profit/loss. So, losing a couple times in a row will definitely not increase your probability of success.

If you're deadset on investigating this (and you probably should just to satisfy your curiosity if nothing else) then you need to read up on something called "Gambler's Ruin". Here's a great read about Gambler's Ruin as it relates to trading futures.
  • #6
  • Dec 9, 2006 1:16am Dec 9, 2006 1:16am
Let me share some of my thoughts with you regarding what you are saying...

You are taking a series of trades & saying that the probability of 5 or 6 consecutive losses in a row are very low...Yet, within the realm of every trade, the probability of losing on that particular trade is still 50%...So basically you are saying that it is very unlikely that one can hit this specific 50% of losing probability several times in a row...This is fine...However, it all depends on what you are actually doing...In my opinion, for what you are saying to be true, you must always trade a pair only from 1 side, either long or short, just to fix 1 of the lots of variables inside this life...Second you need to find entry & exit points...Since your stop loss & profit targets are already set to 100 pips, then your exit is already defined...This leaves up with the entry as the only variable that is not yet defined...

You will also have to chose a pair in which the 100 pips mentioned doesn't represt a small amount of the average daily range...For example, the AUD/USD, a 100 pips there is a big deal, while if the same numbers are applied to the GBP/USD, it might be a disaster when it comes to real life trading...

Now, let's see how likely such a strategy might do in different market conditions, i.e. trending, sideways & directionless or volatile...

This would heavily depend on your fixed side of the market...If we are already set to take our trades as shorts & our pair is in an uptrend, then I think we would get hit very soon...And the small probability of getting 5 or 6 losers in a row will become a fact of life...However, if we are trading from the long side while we are in an uptrend, then our trading would be extremely profitable UNTIL the trend reverses to down & we would get hit once again only to give back almost all of our previous gains...

Now in a sideways market, you would be doing marginally well if and only if the range of the sideways actions is 250 pips or more...That is if you entered your trades in the upper half as shorts or in the lower half as longs, you would still have winning trades...If the range of the sideways action is less than 200 pips, your probability of getting losers in a row is getting higher, and if the range is 100 pips or less, you are not going to make any profits until the sideways period ends...

In directionless volatile periods where ranges of the pair would be 300 to 500 pips without a direction (No trend and no defined sideways range, like the pair moves 200 pips up, then goes 350 pips down, then moves 500 pips up and so on), most probably we will end out breaking even...As we are already trading the market from one side only, so we might win our first trade, lose our second, win our third on double the position size & lose our fourth on a standard position size and then the volatile period fades & the market gets back to normal....

This needs some further statistical investigation integrated with real testing even on the historic data...I think this would be very easy to do during the weekend


Thanks,

Nader
  • #7
  • Dec 9, 2006 1:52am Dec 9, 2006 1:52am
Quoting Leugimp
I believe this is a martingale strategy, doubling up to get back loss plus profit. Whatever you want to call it here is one problem. Your math basically states that the chances of x happening are y, and y being a very small %. In your example you use 1000 coin flips or trades. The problem with this system is that those 9 times out of 1000 will wipe you out. You state that you must use good money management, however your money management has already been defined. - Trade it all the time and 1 to 1 risk to reward(although you did forget the zero and double zero known as the spread). Lastly how about the house limit. Can you trade 1032 lots, and if you lose can you trade 2064, and if you lose again, will you be able to trade 4128 lots. Howabout slippage with these large plays? Can you ever realistically move up in lot size. All of these factors make this an unfavorable game and the only way to win an unfavorable game is to bet big and stop playing when you win. If you plan on trading for a living, you will experience 9 losses in a row using this system, maybee more. This string of losses may not be today nor tomorrow, but it will come if you strictly follow this, and when it does come, you will get wiped out. Not trying to harp on your system, just some concerns.
I thought martingale involved "averaging" your price up or down in a trade so that when price finally retraced and moved the direction you want, you can exit at a price lower/higher that is easier to achieve and exit with profit.

Doesn't matter though. We're just talking about this strategy now, martingale or not.

In regards to your points. The time that you lose 9 times in a row could wipe you out if you don't use sound money management. What if my account is $50,000 and I am still using .01 as my pip value. I'm not sure about other brokers, but with GFT I signed a sheet to adjust my account such that I can have a standard accounr with all the good features, but still trade with minilots and higher leverage like a mini account. So what matters is that you trade to be able to survive the 9 trade loss. This means that you should not trade your account in a manner where it losing 9 in a row would call upon having used over half your account. There should always be cash in your account that you don't use which is only there to provide stability in your margin.

Will I be able to trade 2064 lots? Again that is up to my money management. My broker allows me to, so why shouldn't I be able to if I'm smart about it? In fact my order entry box has a quick button to trade 1 million lots. No lie. I'm sure a lot of brokers do. Trading 5000 lots would be no big deal for the brokers and no big deal for me so long as my account and money management can handle it.

Eventually I do believe you can move up in account size. We are trading with risk every day. There is always some gamble. At some point if this system were to profit we would have to say, "ok, my system is based on the fact that eventually I will win and that will happen sooner than later. Why am I afraid of increasing my lot size because if I lose 9 times in a row I will lose my account? Probabilities show that this is an extremely small possibility. Perhaps although I am not ready to take on a 9 loss sweep in my account, but I will risk my account to bet it won't happen."

That sounds bad because of the betting terminology, but thats what it has to be sometimes. Each trade we make is a bet which we judge in our heads calculating risk/reward, percentages and then if we think it is worth chancing.

Slippage, news and all that will have to be factored in somehow. Its just part of the game. Something to consider is that this could be played differently than a typical grid system. There is only one trade occuring. If you're afraid of slippage, say around news, then you can easily exit the one trade and wait for less volitile times. Something to consider.
  • #8
  • Dec 9, 2006 2:07am Dec 9, 2006 2:07am
Quoting tesla
What a coincidence! I was talking to someone just two days back about the same sort of thing.



It's quite possible you're aware of this already, I'm just going to mention it because the wording you used above sets off warning bells in my head:

Each trade is a distinct entity. Losing on three trades in a row does not make you more likely to win on the fourth. That fourth trade will still have a 50/50 probability of profit/loss. So, losing a couple times in a row will definitely not increase your probability of success.

If you're deadset on investigating this (and you probably should just to satisfy your curiosity if nothing else) then you need to read up on something called "Gambler's Ruin". Here's a great read about Gambler's Ruin as it relates to trading futures.
I will try to find an article about probabilities to prove my point later, but I'll try to explain it now.

I agree each trade, when viewed independently, has a 50/50 chance. As soon as we trade them in succession and want to know the probability of them occuring in a specific pattern then we must make them dependent on one another.

Having six losses in a row has the same probability of having the pattern: win, win, loss, win , loss , loss. For that pattern to occur to completion do you know how many possibilities before it are present? Thousands. Thats the idea. As soon as you take two independent events and correlate them, their probabilities go down drastically. Just flip a coin to see the math in reality.

What're the chances of getting tails on the first flip and then tails on the second flip? You can have tails heads, heads tails, heads heads, tails tails. The probabilty of getting heads or tails on a single flip is always 50%. But when you want to know the probability of a pattern it's no longer 50%. Since we have two tosses the probability is .5 X .5 = .25 or 25%. This is true since there were four possible scenerios in the coin tossing and only 1 out of 4 produced the results we were looking for.

Would you agree that each and every coin flip is its own distinct occurance independent of the previous and next coin flip? Somehow, the probability of getting two losses in a row is only 25% though. That is how my idea for the system will work. In the end through money management, we will play the law of large numbers to win. Thats how the casinos win. They survive the losses and bank on the wins. That's all the doubling down idea is doing. We survive the losses to win big on the winners and erase the losses.
  • #9
  • Dec 9, 2006 2:29am Dec 9, 2006 2:29am
regardless of your account size, if you want to sum up your statistical model, its best outcome will be break even at some point or another unless you find a way to get that edge on your side. you will gather many wins multiplied by the probability that each will occur and losses multiplied by their probability. Although occuring less often, you losing 9 or ten in a row, if risk/reward is 1:1 and you never increase lot value, will mostly wipe out all of your gains. In a perfect world, if the worst case scenario happened in trades 992-1000, and there wasn't a spread or slippage, you would be break even at that point, and by that I mean that every possible combination of wins losses occured in exact % of its probability within those 1000 trades. Expected return would still be zero at some point unless you just quit when you were up. According to your strategy though, you would keep doubling so hypothetically, you could be break even and risking 5000 lots on your next trade on a coin flip. That doesn't sound like good money management to me. what do you do at that point, start over. I hate to do this, but do you think that you are the first to think of this. You really need to question your motivations for trading. Are you trading or gambling, is there a difference to you between the two? What is so different between what you are saying and betting red on roulette. Sounds improbable but it happens almost every night I have been to the casino where I see a table with over 15 blacks in a row. And that is a fair game or rational one. I don't know who said it but do not learn this lesson the hard way - markets can be irrational alot longer than you can remain solvent.
Man who scratches ass should not bite fingernails
  • #10
  • Edited 3:27am Dec 9, 2006 2:33am | Edited 3:27am
hi
firstly im NOT a mathematician
from this page
http://pages.stern.nyu.edu/~churvich/Undergrad/Handouts4/13-Streak.pdf


i calculated (hope theyre right) that in a 1000trade sample with a 50% win rate. the chance of at least one streak of 10 trades (either winning or losing)is 48.4%.
the chance of at least one streak of 15 trades is about 1.5%

the formula is:
probability of at least one streak in the sample = (1+(n-r)*p) * (q ^r)

where, n is the sample size. r is the number of loses in a row. p is winning prob. q is losing probability.


so,
(1+(1000-10)*0.5) * (0.5^10)= 0.484 or 48.4% chance of at least one streak of 10trades in a trade sample of 1000trades. of course as
Taleb cautions there could be more than one streak in that sample

hope i read the formulas correctly. as im NOT a mathematician.
what im interested in , is does anyone have the formula to know the percentile chance for a NUMBER of streaks of a certain length in a trade sample. as there can be more than one in any trading sample.

this is my jun06 post regarding stats and trading
http://strategybuilderfx.com/showpos...&postcount=872
another INTP
  • #11
  • Dec 9, 2006 2:35am Dec 9, 2006 2:35am
Quoting narafa
Let me share some of my thoughts with you regarding what you are saying...

You are taking a series of trades & saying that the probability of 5 or 6 consecutive losses in a row are very low...Yet, within the realm of every trade, the probability of losing on that particular trade is still 50%...So basically you are saying that it is very unlikely that one can hit this specific 50% of losing probability several times in a row...This is fine...However, it all depends on what you are actually doing...In my opinion, for what you are saying to be true, you must always trade a pair only from 1 side, either long or short, just to fix 1 of the lots of variables inside this life...Second you need to find entry & exit points...Since your stop loss & profit targets are already set to 100 pips, then your exit is already defined...This leaves up with the entry as the only variable that is not yet defined...

You will also have to chose a pair in which the 100 pips mentioned doesn't represt a small amount of the average daily range...For example, the AUD/USD, a 100 pips there is a big deal, while if the same numbers are applied to the GBP/USD, it might be a disaster when it comes to real life trading...

Now, let's see how likely such a strategy might do in different market conditions, i.e. trending, sideways & directionless or volatile...

This would heavily depend on your fixed side of the market...If we are already set to take our trades as shorts & our pair is in an uptrend, then I think we would get hit very soon...And the small probability of getting 5 or 6 losers in a row will become a fact of life...However, if we are trading from the long side while we are in an uptrend, then our trading would be extremely profitable UNTIL the trend reverses to down & we would get hit once again only to give back almost all of our previous gains...

Now in a sideways market, you would be doing marginally well if and only if the range of the sideways actions is 250 pips or more...That is if you entered your trades in the upper half as shorts or in the lower half as longs, you would still have winning trades...If the range of the sideways action is less than 200 pips, your probability of getting losers in a row is getting higher, and if the range is 100 pips or less, you are not going to make any profits until the sideways period ends...

In directionless volatile periods where ranges of the pair would be 300 to 500 pips without a direction (No trend and no defined sideways range, like the pair moves 200 pips up, then goes 350 pips down, then moves 500 pips up and so on), most probably we will end out breaking even...As we are already trading the market from one side only, so we might win our first trade, lose our second, win our third on double the position size & lose our fourth on a standard position size and then the volatile period fades & the market gets back to normal....

This needs some further statistical investigation integrated with real testing even on the historic data...I think this would be very easy to do during the weekend


Thanks,

Nader
Please refer to my previous post about how when we correlate two independent probabilities into a pattern it reduces the probability drastically from 50%.

I was only using 100 TP and 100 SL as an example to keep the math simple for now. In reality I think this would work great on an intraday grid strategy that is always in the market. All we need to find is a good SL/TP level that is small enough to enter a trade when there is still enough steam left in the move and large enough that it doesn't get whipsawed in and out from the general trading noise. I'm thinking somewhere from 15-20.

I'd rather keep the discussion on theory right now though. No point in looking at numbers if we don't know what we're looking for.

I don't understand why you say it would only work if it placed only long or short trades and not both. I don't think we can limit ourselves to only one type of trade. The whole basis of this system is that there is a 50/50 shot of where price will go. Up or Down. What is the point of limiting ourselves to only one of the two? In essence you're using the idea of the system against itself.

You want to use only 50% of the possible outcomes to be right 50% of the time. That means you're really only going to be right 25% of the time.

I've been discussing mechanical breakout strategies on the other thread for some time now. So to hint at where I think this strategy will work is the "always in" grid strategy. Say price is at 1.9000. We're playing a 20 SL and 20 TP. Price drops to 1.8980 and triggers a short sell. Why a short sell? Because price was coming from above the 1.8980 trigger. I believe in doing something as simple as this we give ourselves a greater than 50% chance of being successful since the price is most likely to carry the momentum. This also means on trending markets or sharp rises we are guaranteed the wins since the system correctly predicts the movement based on the direction. So I don't know why we'd need to fix the system to only make long or short trades.

Again, my initial 100 TP and SL was only to give simple ideas that for this to work we need to keep the risk:reward equal, if not favored towards the TP side. I don't want to mess with the probability though.

Great responses thus far! Matt
  • #12
  • Dec 9, 2006 2:36am Dec 9, 2006 2:36am
I also may understand that my math is off, you don't need to focus on that, what I am saying is that at some point you may be at break even and risking 5 thousand lots or even 10000 after 1 more loss. at .01 per pip that could amount to a 10,000 loss on a 50K acct. and if you lose that one you will be risking 20K on a 40K acct. One more loss and you do not even have enough money to properly trade your strategy, meaning any attempt to will mean a lower risk:reward and high probability of account ruin. I strongly suggest you avoid this strategy the way it is, but ultimately it is your money, if you want to gamble it is up to you. WHo knows, you might get lucky - and dont kid yourself, if you still have any money after a year or two of trading like this, you will be very lucky.
Man who scratches ass should not bite fingernails
  • #13
  • Dec 9, 2006 3:07am Dec 9, 2006 3:07am
Quoting Leugimp
regardless of your account size, if you want to sum up your statistical model, its best outcome will be break even at some point or another unless you find a way to get that edge on your side. you will gather many wins multiplied by the probability that each will occur and losses multiplied by their probability. Although occuring less often, you losing 9 or ten in a row, if risk/reward is 1:1 and you never increase lot value, will mostly wipe out all of your gains. In a perfect world, if the worst case scenario happened in trades 992-1000, and there wasn't a spread or slippage, you would be break even at that point, and by that I mean that every possible combination of wins losses occured in exact % of its probability within those 1000 trades. Expected return would still be zero at some point unless you just quit when you were up. According to your strategy though, you would keep doubling so hypothetically, you could be break even and risking 5000 lots on your next trade on a coin flip. That doesn't sound like good money management to me. what do you do at that point, start over. I hate to do this, but do you think that you are the first to think of this. You really need to question your motivations for trading. Are you trading or gambling, is there a difference to you between the two? What is so different between what you are saying and betting red on roulette. Sounds improbable but it happens almost every night I have been to the casino where I see a table with over 15 blacks in a row. And that is a fair game or rational one. I don't know who said it but do not learn this lesson the hard way - markets can be irrational alot longer than you can remain solvent.
You'll have to give me some examples in math. I don't see how I would end up break even. Let's say we do 10 trades. 50% lose, 50% win.

First example:
First trade loses and second trade wins. Repeat this 4 more times to get 10 trades. On the first trade we risked 1 lot and 2 on the second trade. Third trade we risk 1 lot, fourth trade 2. Again, repeat.

So we lose 100 on the first trade of each pair, and profit 200 on the second trade of each pair.
The formula is then -100(5)+200(5) = -500 + 1000 = +500 profit

Now let's say we lose the first five trades in a row and win on the 6th. We would risk the following lot amounts. 1-2-4-8-16-32-1-1-1-1.
So we would lose -100 for each lot in the first five trades.
-100 + -200 + -400 + -800 + - 1600 = - 3100.

On the winning trades we would make 32(100) + 1(100) + 1(100) + 1(100) + 1(100). So gains would be 3200+400 = 3600.

Profits would be 3600 - 3100 = +500.

The only possible scenerio in which you would lose is if all the trades that win are in the very beginning and all the trades that lose are at the very end of your trading. This is a perpetual system though since it uses the law of large numbers. Why would you stop trading at 999 trades, or just 57? It's meant to survive the long haul. Even then, what are the chances you limit yourself to 1000 trades and the last 9 are all losers? It would be just like the coin flip example. The probability that the previous 991 trades worked out perfectly such that only losers were left is almost zero.

Yes it is a coin flip. That's all it is. Except since I know I'll be able to survive to the next coin flip, and the next, and the next. Then I know I'll be ok because the coin flip will always work out to nearly 50% in the long run. Hell, the system could only be 1% correct so long as I was able to survive till the end to play that one winning trade. Our chances are better then that though.

No I know I'm not the first to think of something like this. I would feel ashamed for us as traders for not exploring something so simple. Why doesn't it work though? Why aren't people using this? If you keep putting money into the system, and using severe money management, don't the mathematical principles guarantee victory?

Like I've already said. Yes it is betting. Yes it is gambling. Every trader is a gambler. You're betting that through your research the chances of the price going in your favor is good enough that you'll put your money down to make more money. It's as simple as that. Aren't there some brokers with rules specific to certain religions because they view trading as immoral and therefore need special guidelines to trade under their religion?

My motivations for trading are clear. To make money. That's what you're here for too. There is more than one way up a mountain. I've looked at fundamental analysis. I've looked at technical analysis. I've read books and books on trading. I traded the stock market. Forex is different. Currencies do not go up or down forever. If you can survive the long periods of drastic trading then you'll be fine.

I don't know how you can say I'm gambling when I've state several times that this strategy should only be traded with severe margin requirements and micro accounts of .01 a pip; at least to start.
  • #14
  • Dec 9, 2006 3:12am Dec 9, 2006 3:12am
Quoting Leugimp
I also may understand that my math is off, you don't need to focus on that, what I am saying is that at some point you may be at break even and risking 5 thousand lots or even 10000 after 1 more loss. at .01 per pip that could amount to a 10,000 loss on a 50K acct. and if you lose that one you will be risking 20K on a 40K acct. One more loss and you do not even have enough money to properly trade your strategy, meaning any attempt to will mean a lower risk:reward and high probability of account ruin. I strongly suggest you avoid this strategy the way it is, but ultimately it is your money, if you want to gamble it is up to you. WHo knows, you might get lucky - and dont kid yourself, if you still have any money after a year or two of trading like this, you will be very lucky.
I don't see why you're focusing on the highly improbable. Depending on account size, margin, and lot size it is possible for someone to survive many losses in a row. It could be 40 losses out of 50 so long as the wins are every so often to break up the losses. The probability of such a catastrophe is nearly zero.

This is how the casinos win is it not? They survive till the end. They know they can take a $500,000 hit on a slot machine because there will be many other people to gamble their money and lose. Is this not the same thing? The house always wins.
  • #15
  • Dec 9, 2006 3:40am Dec 9, 2006 3:40am
Quoting tesla
If you're deadset on investigating this (and you probably should just to satisfy your curiosity if nothing else) then you need to read up on something called "Gambler's Ruin". Here's a great read about Gambler's Ruin as it relates to trading futures.
Read your article and I agree with it. It strengthens what I've been saying. They said the problem with gamblers is they don't have the bankroll of the casino. Gamblers only need one or two bad streaks and they're out of the game, but if they had more money to play and get back to the expected returns they'd be ok. That's why I'm saying this strategy needs to be implemented on a micro lot account in the beginning.

If you can survive the long haul you'll be good. Money management. Slim initial profits. Be patient!
  • #16
  • Dec 9, 2006 4:24am Dec 9, 2006 4:24am
Quoting anthonybaker
hi
firstly im NOT a mathematician
from this page
http://pages.stern.nyu.edu/~churvich/Undergrad/Handouts4/13-Streak.pdf


i calculated (hope theyre right) that in a 1000trade sample with a 50% win rate. the chance of at least one streak of 10 trades (either winning or losing)is 48.4%.
the chance of at least one streak of 15 trades is about 1.5%
Your calculations were correct. In fact, according to those formulas, out of 1000 trades there is a 48% chance of getting a 10 loss losing streak.
What about out of 100 trades? (1+(100-10).5).5^10 = 4.49% chance.

I'll ask the question now then. What do we need to do to make us more than 50%? I think the simple idea of a grid system deciding on long or short signal based on the direction from which price hits a trigger can greatly affect results. If it brings our success rate to just 60% then out of 1000 trades we can expect to have a 10 trade losing streak of:
(1+(1000-10).6).4^10 = 6.23%. See what that did? Just a little bit more accuracy greatly affects the possibilities. It brought it down from 48% to just 6%.

In 100 trades with 60% success a 10 loss streak has a .57% chance of happening. So then how do we trade? I would suggest, start with a micro account, high leverage, and a balance larger than needed to open a micro account. Still trade with just .01 pip values. Even if you lose 10 trades in a row and need to trade 512 times your initial trade value that is still only a value of 5.12 per pip, not even a standard lot yet. So with a few thousand I think it's possible. If you have $20K I think you've very safe.

Play the market and see how things go. How well does it trade. Is it 60% accurate? Is it only 50%? Go from there. Make a decision. Would you feel confident enough to triple down on the losses such that every trade ended up making you profit eventually essentially. That is when you would hope for consecutive losses and a win to multiply the profits. That would be a tough decision to make considering the risks though.

The point I'm trying to make is to play like the casino. Be the house and the market maker/broker. What is their game? They know 95% of us will lose and they can take losses on the 5% that win. Are they reading charts and plotting trendlines? I doubt it. They have a computer counter balancing our trades against others knowing that most of us will lose. Their success rate is 95%. Ours is 50% if we play by their rules. How do we play with their formula to trade individually? I think that is where this thread should eventually go. What are ways to institute house rules on our own turf?

I'll point out that I agree with Leugimp that it can be very easy to wipe out an account. That is why I have been stressing money management from the very beginning. This is a severely difficult way to trade. But, I don't think it would be impossible to not trade this way. If it's impossible then casinos would have gone out of business years ago.

The article posted by anthonybaker is an excellent one providing the formulas for finding probability of losing streaks in x amount of trades. There was also another point that was brought up in the same article. It mentioned why people will lose when doubling down on roullete. It noted that casinos limit their patrons from doubling down more than 10 times in a row. Why? The probability of the person losing that 11th time is too small for the casino to risk having to pay out so much money. The casino is protecting itself from why I think this system will work. A 10 loss losing streak is 48% likely in 1000 trades. An 11 loss losing streak is 24% likely. The probabilities begin to plummet drastically the longer you trade. Eventually you will win. It all depends on your bank account and money management.
  • #17
  • Edited 9:07am Dec 9, 2006 6:49am | Edited 9:07am
ive just been reading this interesting article regarding the nonnormal nature of market returns. a normal return of distributions looks like the classic bell curve. as i was trying to find the formula for the distribution, the percentage chance for a number of streaks of a certain length in a trade sample.
eg. 50% chance of 6 streaks of 10 consecutive losing trades in a 1000 trade sample with a win rate of 50% or 30% chance of 10 streaks of 10 consecutive losing trades in a 1000 trade sample with a win rate of 50%.

http://www.billakanodoodahs.com/2006...-golden-swans/

this link explains the terms
http://www.stat.tamu.edu/stat30x/notes/node3.html


although his work is regarding the es, it showed that black and gold swans ie major daily drops or rises of the es by 3%, were four times more likely to happen than normal distribution stats showed

so you are correct to caution leaving extra padding/ money to ride out drawdowns
another INTP
  • #18
  • Dec 9, 2006 12:54pm Dec 9, 2006 12:54pm
Not sure I understand what your link is explaining in regards to black swans and such. Was he measuring how many up days or down days there were in the S&P500?

I'm just wondering if how people would actually trade really could hit 10 losses in a row more often than what these statistics would say. Because in that article I get the feeling it was analyzing the statistics of always betting long or always betting short. How accurate are discretionary systems though? Like I said, how long would the market let you be right about being wrong? An "always in" system would almost be necessary then. It might also require either scalping for only a few pips or waiting for large movements of 150 or more. Price rarely stays in a 150 range for very long. Safeguards such as how we construct the system could protect us further.

I'm also wondering about the double down line strategy. We allow our losses to build up massively through the doubling down so that we can profit of just the one win. Is our risk:reward then skewed towards the loss side even though we win always? Maybe one other way this would survive is to triple down or use some other multiplied factor besides 2.

so it would be: 1 -2 - 6 - 18 - 54 - 162. This way if we assume we'll get streaks of 3-5 losses in a row often it makes our winners VERY profitable. If we have five losers in a row and win on the 6th we would have made profit as if we made 108 winners in a row. Get what I'm saying? Our profit curve goes up massively. By taking those big wins when we can, i.e. very often assuming the chance of a long losing streak is not high, then by the time we hit that long losing streak it is going to be fighting against our highly enlarged account. It would be a gamble in the beginning, but if we don't get greedy and not step up the lot size I think we should be fine.

What does everyone think? Matt

P.S. - anthonybaker - fine articles although I'm going to have to reread them. Mind if you could try explaining them some more?
  • #19
  • Dec 9, 2006 1:46pm Dec 9, 2006 1:46pm
Quoting anthonybaker
ive just been reading this interesting article regarding the nonnormal nature of market returns. a normal return of distributions looks like the classic bell curve. as i was trying to find the formula for the distribution, the percentage chance for a number of streaks of a certain length in a trade sample.
eg. 50% chance of 6 streaks of 10 consecutive losing trades in a 1000 trade sample with a win rate of 50% or 30% chance of 10 streaks of 10 consecutive losing trades in a 1000 trade sample with a win rate of 50%.

http://www.billakanodoodahs.com/2006...-golden-swans/

this link explains the terms
http://www.stat.tamu.edu/stat30x/notes/node3.html


although his work is regarding the es, it showed that black and gold swans ie major daily drops or rises of the es by 3%, were four times more likely to happen than normal distribution stats showed

so you are correct to caution leaving extra padding/ money to ride out drawdowns
Yes it's been proven that the markets don't follow a bell shaped distribution. The reason? The markets are moved by phychology and not mathematics. So this throws all your statistics off.

Secondly even if you're statistics were right, it's very hard emotionally to trade a system just based on statistics because then you're excluding price action. I know, I've tried.
  • #20
  • Dec 9, 2006 1:53pm Dec 9, 2006 1:53pm
last Post. If you lose first trade, you are down 100 pips. If you lose second trade you are cumulatively down 300 pips. After the third 700 pip loss cumulative. Whenever you do hit, you will only be up 100 pips at .01 per lot is a whopping $1.00, no matter when you win. lets say you are up $600 at some point. well now lets see what happens when that 12 loss streak happens, -100,-200,-400,-800,-1600,-3200,-6400,-12800,-25600,-51200, -102400, -204800. Add those all up and you have -409,500 *.01 = -$4,095. Granted if you hit the next one, you will be at $4195 added to your original $600 for a cumulative gain of $700, however, look at the risk on that particualr winning trade. Your account prior to the win will be $46,505(50,000+600-4095). on that particualr trade you are risking almost 10% of your capital. And what if you lose, on the next trade you will have roughly 40,000 and risking 8,000. One more loss and you now have 32,000 and risking 16,000. One more loss and now your system needs to be refunded to go on. I know it is highly improbable, but improbable is not the point. The point is that if you take it as a given that you will trade for the rest of your adult life, you will encounter one of these streaks at some point or another, and unless your pockets are much much deeper than 50K, you will eventually get burned. Think about it, on the trade after the last one described, you will be betting(if you fund your account, 32,000 on a coin flip, 64,000 after that 128,000 after that. You may say that it is improbable to happen, yet over time, it will happen. and when it does, you are in trouble. Remember, every 100 pip win will yield you 1$. Therefore, no matter how you look at it, on win #1000 you will be up a total of $1000.00. Your account will be at $51,000(less spread). Is that $1000 gain enough to really withstand such a losing streak. Only if you have very very very deep pockets. And for what, to be up 1.00 after every win. And how often do you plan on gaining 100 pips. If you were winning every day for a year 100 pips, you would be up around $200 for the year. Is it really worth this risk? Lastly, I would recommend reading Benoit Mandelbrots the Misbehavior of markets. You really need to understand that normal statistics do not work in trading and that events that are 8-9 standard deviation events occur much more often than their normal probabilities would suggest. If you want to take advantage of statistics, you are much better off taking advantage of that one.
Man who scratches ass should not bite fingernails
0 traders viewing now