Stupid System | Forex Factory

Options

Go to
Permalinks
Print Thread

Similar Threads

Stupid Trading System 5 replies

The simple stupid system 855 replies

very stupid question but grrrrrrrrr!!!!!!!!! 2 replies

Call me stupid - Indicator discovery 32 replies

Stupid maths 19 replies

Attachments: Stupid System
Exit Attachments

Stupid System

  • #1
  • First Post: Edited Jul 11, 2007 2:11pm Jul 10, 2007 7:54pm | Edited Jul 11, 2007 2:11pm
PLEASE NOTE: This is a work in progress. I have recently updated the gains listed here based on further analysis.

This approach is ultra simple. It is completely mechanical and uses no charts, indicators, or predictions. It is based only on price action and probabilities. It is not a humongous winner, but the gains appear to be stable and consistent over many years. It seems to be applicable to most pairs.

I have run historical studies from 2003-2007 for the following pairs, with the overall gains listed:

EURUSD (805)
EURCHF (2997)
EURJPY (3435)
EURGBP (1508)
GBPUSD (404)
GBPCHF (2775)
GBPJPY (531)
USDCAD (1087)
USDCHF (-1627)
USDJPY (-7)
AUDUSD (-284)

This system is designed to take small daily profits from a basket of pairs. The individual gains are small, but accumulate over time.

The approach is based on the fact that literally 99% of the time, there is action on BOTH sides of the opening price. Only 1% of the time is the opening price also the high or low of the day. There is a very high probability that the distance from the open to the high or low of the day will be AT LEAST a certain amount.

If you calculate the average amplitude (1/2 daily high-low range) for each pair, you will find that this corresponds to the average open-high range, the average open-low range, and the average open-close range. The amplitude is a kind of basic measure of inherent volatility. That's just how much the price wiggles around (either up or down) every day.

Now, if you take 1/2 the amplitude as a profit target, you will find that the probability of hitting that target from the opening price on any given day (whether long or short) is in most cases 70% or higher.

So, here's how I did my historical studies and how the system works:

1. Use the following six pairs as an example basket: EURUSD, EURCHF, EURJPY, GBPUSD, GBPCHF, GBPJPY

2. For each pair, dump the historical data to a spreadsheet and calculate the profit target as 1/2 the average amplitude for the previous month. This means the profit target changes gradually in response to changes in the volatility of the pair.

3. Determine the "trade side." I did this in the simplest possible way ... by calculating the historical %age of up days and down days. Each pair will have an edge on one side or the other. It so happens that all of these six pairs historically have more up days than down days. That means by default that I will go LONG.

4. At the open of each trading day, go LONG in each of these pairs. Exit at the profit target or if it is not hit, at the close of the day.

5. If the previous day was a loss, do the opposite on the next day (SHORT at the open).

6. Repeat.

This example averages 202 pips per month for the past 5 years. I have not yet worked out an optimal stop, but I will post more later on stops and drawdowns.

Other pairs/combinations will do better. However, having said all this, I would have been really happy to see it work out with every pair, but there are a few that did not, namely: USDCHF, AUDUSD, and USDJPY. There are probably others as well. I have no explanation and haven't thought about it in detail. Perhaps someone has some ideas.

I welcome your comments, criticisms, and improvements!

All the best.
  • #2
  • Jul 10, 2007 8:50pm Jul 10, 2007 8:50pm
Interesting system! Thank you for sharing.

I am trading (demo) very similar system myself. It runs on the propobility as well, very similar approach. I have found that using trailing stops and hard profit targets would increase the profits even more. My system is using RANDOM direction (basically coin flip) to go long or short on day OPEN and it is profitable.

Currently I am working on money management based also on statistics to be added to this system. I am counting max number of consecutive times the system was ever wrong and increase number of lots played accordingly The danger is not to be victim of 'gambler ruin' (or should I say 'trader ruin'). LOL! The only problem is that the random entry is kind of difficult to backtest, so for each settings I repeat test 100 of times to see if values work. (I am using TradeStation 8 to do that).

Your idea to trade few pairs at the same time looks very promising!

Maybe this system is not so stupid, after all !
  • #3
  • Jul 10, 2007 9:05pm Jul 10, 2007 9:05pm
Very interesting, thanks for sharing. I would like to make a question...

Is it necesary to go on the same direccion in all the pairs? For example, what about if i have 5 pairs that are down direction in average but the sixth pair always goes up (EUR USD)... do I have to go short also in this pair?

Thanks for your time

Ado
Ado Whether you believe you can do a thing or not, you are right.
  • #4
  • Jul 10, 2007 11:24pm Jul 10, 2007 11:24pm
Can I borrow your coin ??
If we aren't supposed to eat animals..... why are they made out of meat??
  • #5
  • Edited 1:58pm Jul 11, 2007 4:25am | Edited 1:58pm
Quoting MrGuru
Interesting system! Thank you for sharing.

I am trading (demo) very similar system myself. It runs on the propobility as well, very similar approach. I have found that using trailing stops and hard profit targets would increase the profits even more. My system is using RANDOM direction (basically coin flip) to go long or short on day OPEN and it is profitable.

Currently I am working on money management based also on statistics to be added to this system. I am counting max number of consecutive times the system was ever wrong and increase number of lots played accordingly The danger is not to be victim of 'gambler ruin' (or should I say 'trader ruin'). LOL! The only problem is that the random entry is kind of difficult to backtest, so for each settings I repeat test 100 of times to see if values work. (I am using TradeStation 8 to do that).

Your idea to trade few pairs at the same time looks very promising!

Maybe this system is not so stupid, after all !
Hi Guru, these are interesting ideas. Here are a couple off the top responses:
1. I'm not sure about the random direction. I think we're on the same page in that we are agnostic to the trend. At the same time, it seems that there is a slight statistical edge for any given pair in trading predominantly on one side or the other. In the example set I have given above, if I flip the direction and go short by default, the results are dramatically lower. The other thing that you mentioned is that backtesting randomness can get complicated. For X thousand trading days, you have to generate hundreds of X thousand sequences of random entries (Long/Short). The issue is that random sequences in real life tend to cluster together. There's the classic experiment of the math teacher who gives 2 coins to his class, instructs half the class to flip the coin a hundred times and record the results, instructs the other half to make up a random sequence without flipping the coin, and tells both groups to write their final result on the board. The teacher then leaves the room, and upon returning instantly identifies the sequence based on real coin flips and the sequence that was made up. The reason is that when we try to "make up" a random sequence, people tend to alternate too often, thinking this is "randomness." But in reality, a random sequence of heads and tails will have some long series of heads and long series of tails, which is why the teacher can spot the real random sequence right away. If this is true in your case, then having a series of long entries or short entries at various times could possibly have very different results. This would also give me cause for concern with a money management system based on consecutive wins/losses. For every random historical sequence, you would get different sets of consecutive wins/losses at different times. Even running the strategy simulation hundreds of times, you would end up arbitrarily choosing one "maximum" number of consecutive wins/losses. But what would you gain by doing this? It's probably much easier to just scale position size with account size, constantly adjusting fractional lot sizes up and down so that your equity risk remains constant.

2. The other thought has to do with the trailing stop. For some people, this is kind of a religious topic. The reality is that trailing stops are made for trends. Profit targets are made for ranges. If you keep a trailing stop in tight range, you will always 100% get stopped out. If you use fixed profit targets in trends, you will miss big opportunities. And there lies the rub. You're left with choosing one or the other, using some kind of hybrid approach, scaling in and out, etc. No doubt, having a good exit strategy is of paramount importance. But that strategy has to align with the underlying approach of your system and the behavior of the market you're trading. For that reason, I don't think that using a random entry with a trailing stop will get you very far. You would be ignoring the trend on the entry but using a trend-following exit strategy.

Having said all of that (and I think I've written too much already), I think that the place for real improvement with the idea I've presented here is the exit strategy. That's where I would love for us to explore improvements.

The entries are fairly easy. For example, I have suggested in the basket of pairs above going long on EURUSD every day and taking small profits from the open. This simple strategy wins 71% of the time on EURUSD between 2003-2007 ... 837 times out of 1175 trades. The maximum risk during that time is -265 pips. The average loss (exiting at the close) is 58 pips. The other pairs in the basket have similar characteristics. The point is that while I'm sure the entry points and risk management could be improved, I think the real opportunity to turn this stupid system into a really smart system is with a really smart exit strategy.

That's my two cents.

What do you think?
  • #6
  • Edited 2:34pm Jul 11, 2007 5:00am | Edited 2:34pm
In my last post, I was responding to the idea of random entries. That got me thinking about the value of reversing the direction after a losing day.

So, I went back to my historical studies and took out the reversal rule. That means I only trade one side, always. The results were very interesting.

Here are some updated numbers:

EURUSD (941)
EURCHF (3608)
EURJPY (627)
GBPUSD (405)
GBPCHF (4185)
GBPJPY (-1107)

The most obvious result is that the overall gain of the CHF pairs went way up, while the overall gain of the JPY pairs went down. I'm not really sure why. It could be that this kind of system likes choppier markets. It could be that the strong sustained upward trends of the JPY pairs have something fundamental to do with it. It just seems that frequent reversals on the JPY pairs pay off much more so than with the other pairs. It could be that these pairs exhibit particular intraday price patterns such as: higher opens, downward momentum, higher closes. If I were a dealing desk in control of quotes, that's what I would probably do to take out stops and profit in short term movements. Anyway, I don't really know.

Overall, for the basket presented here, the results are lower and would come out to 160 pips per month if the reversal rule is eliminated. But this depends on which pairs you trade. Other pairs showed big improvements. For example, USDCAD (1087 -> 1270), USDCHF (-1627 -> 555), and GBPCHF (2775 -> 4185).

What does this mean?

The reversal rule seems to impact the pairs in different ways. This suggests some fundamental relation between CHF and JPY pairs, and I don't even pretend to understand what that could be.

It seems like there is more to gain by eliminating the reversal rule, concentrating on the CHF pairs, and ignoring the JPY pairs, because this system doesn't seem to fit the JPY pairs as a group very well anyway.

Either way, the stupid system still delivers small consistent gains over a long time.

Maybe someone else has other thoughts.
  • #7
  • Jul 11, 2007 5:21am Jul 11, 2007 5:21am
Did you take into account the spread?
Did you calculate maximum draw down?
Any idiot can take profit, real traders know how to take losses!
  • #8
  • Jul 11, 2007 5:32am Jul 11, 2007 5:32am
Quoting Inflection

I have run historical studies from 2003-2007 for the following pairs, with the overall gains listed:

EURUSD (1191) MAIN PAIR > BUY
GBPUSD (807) MAIN PAIR > BUY
USDCHF (-1496) CORRELATED PAIR > BUY???
AUDUSD (-427) CORRELATED PAIR > BUY???
Altough I spent just a few seconds on this (I promise I will spend more time on this system...) I see the the negative pips comes from correlated pairs.

IF EUR/USD is heading upwards, it's clear that USD/CHF will be heading downwards, so by BUYING EURUSD, you must SELL USD/CHF because of correlation.

http://www.forexfactory.com/showthread.php?p=1489664

I hope this will help...

Happy trading

P.S. I need to get some friggin' sleep, cause EUR shook me all nite long Will get back here later
  • #9
  • Edited 7:25am Jul 11, 2007 5:54am | Edited 7:25am
Quoting Ado
Very interesting, thanks for sharing. I would like to make a question...

Is it necesary to go on the same direccion in all the pairs? For example, what about if i have 5 pairs that are down direction in average but the sixth pair always goes up (EUR USD)... do I have to go short also in this pair?

Thanks for your time

Ado
First of all, keep in mind that I'm a newcomer to Forex. I'm still in my gray zone with just about everything.

With that caveat, I would say it is not necessary to go the same direction with all the pairs. It just so happened that the six pairs I picked for a "basket" happened to fall on the LONG side. I just picked them because they had good overall results and because they had the symmetry of crossing the Euro and Pound against the other major currencies. I would think that the important thing is to be on the side with the higher probability. For USDCAD, for example, that turns out to be the short side.

The question is whether or not I've come up with a useful way to identify the "trade side." I took a very simplistic approach by using the %age of up days and down days over a long period of time. The percentages are small (e.g. 52% - 56%), but in every case except one, this ended up being the side with the higher returns in this particular system. This could be pure coincidence. But I could cross check it with other measures to see if there is confirmation.

Anyhow, the one exception in my historical studies was EURGBP. It has higher returns on the long side, but it has a slightly higher number of down days than up days. But this number is very close to 50%, so I took the liberty to choose the side with discretion.

Now, somebody is bound to bring up the issue of correlation between currency pairs. Here's where things get muddy.

In the example basket I posted here, I have included pairs which have a very high negative correlation, such as EURUSD and GBPCHF. The correlation between these is something like -95%. So, it is mathematically true that these pairs move in opposite directions. But does that mean in this particular system that if I go long both pairs, that one negates the other? In short time frames, probably not.

In fact, for these two pairs in this system over more than 1100 trade days they both win or lose TOGETHER more than 60% of the time. The other days one wins and the other loses.

I'm not sure that correlation figures will have much to do with the performance of this system.

Does anyone else have other thoughts?
  • #10
  • Edited 2:35pm Jul 11, 2007 6:24am | Edited 2:35pm
Quoting Dexis
Altough I spent just a few seconds on this (I promise I will spend more time on this system...) I see the the negative pips comes from correlated pairs.

IF EUR/USD is heading upwards, it's clear that USD/CHF will be heading downwards, so by BUYING EURUSD, you must SELL USD/CHF because of correlation.

http://www.forexfactory.com/showthread.php?p=1489664

I hope this will help...

Happy trading

P.S. I need to get some friggin' sleep, cause EUR shook me all nite long Will get back here later
Dexis, thanks for your feedback. I look forward to your input.

If you look at the updated numbers where I consider the system without the reversal rule, then USDCHF goes positive in a big way. As I mentioned in another post, the idea of eliminating the reversal rule dramatically improves all the major CHF pairs: USDCHF, EURCHF, and GBPCHF.

Also, if you look at the historical results in this system and consider every trading day EURUSD against USDCHF, you will find that in more than 1100 trading days both pairs win or lose TOGETHER ON THE SAME DAY more than 60% of the time. The other days one wins and the other loses.

I mentioned this in another post where I compared EURUSD and GBPCHF (with -95% daily correlation) and came up with exactly the same results.

I'm not sure that correlation numbers will really play a big factor in the performance of this kind of system.

As I said, I'm a newcomer to Forex so I'm still in my gray zone with a lot of this stuff, but I wonder if talking about correlation figures is a lot like talking about inflation. The inflation rate is the bedrock of every retirement plan pitch from every mutual fund company in the world, but do we really feel the impact of a fractional %age of inflation? Okay, maybe if you're buying milk, eggs, cheese, and whatever else is in the hypothetical inflation basket of consumer goods ... but think about cars, gasoline, or laptop computers. The prices of these haven't changed for years and the price appreciation of things like gasoline have definitely not kept up with inflation. So what gives?

Correlation is a fact. Don't get me wrong about that. But I think the impact of correlation depends on the timeframe and the specific system. This isn't a buy-and-hold approach, so I'm just not convinced that correlation will be a major impact.

Anyhow, I should learn more about this stuff, so I'll read up. Thanks for the link.

Let me know what you think.
  • #11
  • Edited 2:44pm Jul 11, 2007 7:19am | Edited 2:44pm
Quoting bmchunu
Did you take into account the spread?
Did you calculate maximum draw down?
This is a great question. Always the first to be asked with any new system posted, but of course, always necessary to consider. The spread is obviously inescapable and for a system with small profit targets, the spread is a major impact. 4 pips on a system with a 20-40pip expectation is a big bite.

The good news is, yes. While I did not directly factor the spread into each transaction, the profit targets I used in my calculations are all rounded down. That means there's plenty of room for the extra pips to cover the spread.

As a sanity check, I quickly plugged in the numbers for the GBP pairs with spreads between 2.5 to 4 pips. No problem.

In fact, I should mention this as well. The idea of taking 1/2 the average amplitude as the profit target is simple, but not optimal. Actually, almost every pair I tested has room for some extra pips profit on top of the spread. This can have a really big overall impact.

Anyway, I haven't really optimized anything with this system. My goal is to find something simple, robust, and mechanical.

It would be great to have help from you guys in this regard.

As far as the drawdown goes, that's another question. As I originally conceived this idea, I had expected to simply use a single catastrophic stop loss that would be large enough to not be hit too frequently and small enough to actually protect my account and eliminate the really gigantic losses. I'll write more on this later, because I think that there might be something to using the max drawdown for each pair as a risk/reward ratio that could help select the best basket to use. Then I'll look at the impact of various stop strategies.

I will also write up my money management and position sizing numbers in a separate post. This will show how much room there is before a margin call would come my way. I think this is what you're after in terms the overall maximum drawdown.
  • #12
  • Jul 11, 2007 7:52am Jul 11, 2007 7:52am
I like the way you think.

Figuring out the major trend direction and taking trades only in that direction is a sensible approach. You'll find it's a recurring theme.
You are in a maze of twisty little passages, all alike.
  • #13
  • Jul 11, 2007 8:30am Jul 11, 2007 8:30am
The reason I used random entry is to make backtesting as random as possible. I agree with you that this is not the way to trade for real and better enter strategy is needed. Since the whole thing is based on statistics it is actually working! It should not matter what direction you going. If you have an edge and make large number of trades this system must work.

As far as stop loss, I am counting to catch big waves early. I do not close position at closing. Every trade must hit the trailing stop or hard target. Then I would open new position on next day open. Sometimes trade takes few days. In that case it makes sense to use trailing stop.

I think we are on the same page as far as principles of this system and we can learn few things by comparing our work.

I will try to post some results from my system in the evening (central time).
  • #14
  • Jul 11, 2007 9:10am Jul 11, 2007 9:10am
Here's a coin tossing system I found a while back:

1. Toss a coin for long or short entry. Assume long for the remaining steps.
2. If the entry bar closes higher (or lower for short) than the preceding bar, keep the position open until the next bar closes. If the entry bar closes lower than entry, close immediately.
3. If the current bar closes higher (or lower for short) than the preceding bar, keep the position open.
4. Repeat #3 until a bar closes lower than the close of the previous bar.

You will make money over time. Why? Because you're long or short an equal number of times and you cut your losses short and let your profits run.

Add a statistical bias so you're long or short when you're "supposed" to be and the result improve further.

No TA, no oscillators, no Elliott Waves.
You are in a maze of twisty little passages, all alike.
  • #15
  • Jul 11, 2007 9:11am Jul 11, 2007 9:11am
Quoting Inflection
In my last post, I was responding to the idea of random entries. That got me thinking about the value of reversing the direction after a losing day.

So, I went back to my historical studies and took out the reversal rule. That means I only trade one side, always. The results were very interesting.

Here are some updated numbers:

EURUSD (1047)
EURCHF (3608)
EURJPY (1026)
GBPUSD (1038)
GBPCHF (5656)
GBPJPY (1356)

The most obvious result is that the overall gain of the CHF pairs went way up, while the overall gain of the JPY pairs went down. I'm not really sure why. It could be that this kind of system likes choppier markets. It could be that the strong sustained upward trends of the JPY pairs have something fundamental to do with it. It just seems that frequent reversals on the JPY pairs pay off much more so than with the other pairs. It could be that these pairs exhibit particular intraday price patterns such as: higher opens, downward momentum, higher closes. If I were a dealing desk in control of quotes, that's what I would probably do to take out stops and profit in short term movements. Anyway, I don't really know.

Overall, for the basket presented here, the results are lower and would come out to 254 pips per month if the reversal rule is eliminated. However, the overall results are actually slightly higher across all the pairs I tested including AUDUSD, USDCHF, USDJPY, etc. and particularly with USDCHF (from -1300 pips to more than +1000) and GBPCHF (from almost 4000 pips to more than 5500 pips)!

What does this mean?

It suggests to me that eliminating the reversal rule makes the system simpler and more robust. That means it might work across more markets in more timeframes. It also suggests some fundamental relation between CHF and JPY pairs, but I don't even pretend to understand what that could be.

Either way, the stupid system still delivers small consistent gains over a long time.

Maybe someone else has other thoughts.
Please what were you take profit for each of the listed pairs?
  • #16
  • Jul 11, 2007 10:31am Jul 11, 2007 10:31am
this look similiar to http://www.forexfactory.com/images/misc/paperclip.gif Part Time Wealth Building Trading System - Daily Scalping
they also play enter on open and close at the end of the day. may be you can get some info you can use to modified your trade.

Quoting Inflection
This approach is ultra simple. It is completely mechanical and uses no charts, indicators, or predictions. It is based only on price action and probabilities. It is not a humongous winner, but the gains appear to be stable and consistent over many years. It seems to be applicable to (almost) every pair.

I have run historical studies from 2003-2007 for the following pairs, with the overall gains listed:

EURUSD (1191)
EURCHF (2957)
EURJPY (3600)
EURGBP (1502)
GBPUSD (807)
GBPCHF (3896)
GBPJPY (2728)
USDCAD (1071)
USDCHF (-1496)
USDJPY (49)
AUDUSD (-427)

This system is designed to take small daily profits from a basket of pairs. The individual gains are small, but accumulate over time.

The approach is based on the fact that literally 99% of the time, there is action on BOTH sides of the opening price. Only 1% of the time is the opening price also the high or low of the day. There is a very high probability that the distance from the open to the high or low of the day will be AT LEAST a certain amount.

If you calculate the average amplitude (1/2 daily high-low range) for each pair, you will find that this corresponds to the average open-high range, the average open-low range, and the average open-close range. The amplitude is a kind of basic measure of inherent volatility. That's just how much the price wiggles around (either up or down) every day.

Now, if you take 1/2 the amplitude as a profit target, you will find that the probability of hitting that target from the opening price on any given day (whether long or short) is in most cases 70% or higher.

So, here's how I did my historical studies and how the system works:

1. Use the following six pairs as an example basket: EURUSD, EURCHF, EURJPY, GBPUSD, GBPCHF, GBPJPY

2. For each pair, dump the historical data to a spreadsheet and calculate the profit target as 1/2 the average amplitude for the previous month. This means the profit target changes gradually in response to changes in the volatility of the pair.

3. Determine the "trade side." I did this in the simplest possible way ... by calculating the historical %age of up days and down days. Each pair will have an edge on one side or the other. It so happens that all of these six pairs historically have more up days than down days. That means by default that I will go LONG.

4. At the open of each trading day, go LONG in each of these pairs. Exit at the profit target or if it is not hit, at the close of the day. Use a catastrophic stop loss, such as 200 pips. Trade small lot sizes. This will eliminate gigantic losses and will increase the overall return.

5. If the previous day was a loss, do the opposite on the next day (SHORT at the open).

6. Repeat.

This example averages 281 pips per month for the past 5 years.

Other pairs/combinations will do better. However, having said all this, I would have been really happy to see it work out with every pair, but there are a few that did not, namely: USDCHF, AUDUSD, and USDJPY. There are probably others as well. I have no explanation and haven't thought about it in detail. Perhaps someone has some ideas.

I welcome your comments, criticisms, and improvements!

All the best.

EDIT: In another post, I have considered the scenario of eliminating the reversal rule (#5). The outcome is interesting and the overall results are positive. Please read further.
  • #17
  • Edited 11:45am Jul 11, 2007 11:12am | Edited 11:45am
Quoting stevegee58
Here's a coin tossing system I found a while back.......
Hi Stevegee

Have you tested this, and if so, over what time frames?

Sorry to be so negative, but it sounds too good to be true. If trading was this easy, we'd all be millionaires.

If it's possible to trade profitably on the basis of a coin toss entry, then all of the entry systems that have ever been devised are a complete waste of time, as is the maxim "the trend is your friend", since you'd be counter-trend trading, randomly, 50% of the time.

Cutting losses short and letting profits run is profitable only to the extent that markets trend, and far enough to overcome costs.

In all honesty, I'd be delighted to be proved wrong, and I'd be the first to congratulate you.

I admit that I'm skeptical, but I apologize if this any of this sounds facetious or mocking, I don't intend it to be.

Best wishes
David

[My apologies to Inflection - I didn't mean to hijack your thread. This is my last post on this subject.]

[Edit #2: perhaps I've been too hasty. It's very possible that the best systems are simple. Combine this idea with (1) trading in a long enough timeframe to make costs less significant (and perhaps allow positive carry to offset this further), and (2) trading only in the direction of the trend in an even longer timeframe, and I suspect the probability improves significantly. But of course that's no longer a truly random entry.....]
  • #18
  • Jul 11, 2007 2:56pm Jul 11, 2007 2:56pm
Here are the profit target ranges over the period of this study [with averages in brackets]:

EURUSD 15-41 [25]
EURCHF 9-22 [13]
EURJPY 16-48 [28]
EURGBP 5-17 [10]
GBPUSD 20-58 [35]
GBPCHF 23-62 [38]
GBPJPY 24-70 [42]
USDCAD 16-40 [25]
USDCHF 15-46 [30]
USDJPY 15-35 [24]
AUDUSD 10-28 [17]
  • #19
  • Edited Jul 12, 2007 1:57am Jul 11, 2007 4:57pm | Edited Jul 12, 2007 1:57am
Quoting stevegee58
Here's a coin tossing system I found a while back:

1. Toss a coin for long or short entry. Assume long for the remaining steps.
2. If the entry bar closes higher (or lower for short) than the preceding bar, keep the position open until the next bar closes. If the entry bar closes lower than entry, close immediately.
3. If the current bar closes higher (or lower for short) than the preceding bar, keep the position open.
4. Repeat #3 until a bar closes lower than the close of the previous bar.

You will make money over time. Why? Because you're long or short an equal number of times and you cut your losses short and let your profits run.
No

This algorithm is easily computable. Let average move into either direction be N (pips), resulting profit P (pips) and spread S (pips too). Exclude zero moves (they don't change anything), assume that moves down and up are equally probable (1/2).

What happens after first move? In 1/2 cases it goes into false direction and we loss -1/2N pips. In 1/2 cases it goes into favour of us and we are at N pips profit AND in square one - we look at same profit again, shifted by N pips, but having less probabilty (1/2).

Mathematically this looks as follows (we should subtract spread from overall profit only):

P=1/2(P+N)-1/2N-S

Solving that:

2P=P+N-N-2S
P=-2S

Result is sad - statistical profit is negative, twice bigger in value than spread.

If we could use some asymmetrical probabilty (statistical bias), then things are somewhat better (X+Y=1, X>Y, X is winning probabilty):

P=X*(P+N)-Y*N-S
P=PX+NX-NY-S
P(1-X)=N(X-Y)-S
P=(N(X-Y)-S)/(1-X)

This is positive, when N(X-Y)-S>0 or (2*X-1)>S/N or X>(S/N+1)/2. Assume S=3 (good spread), N=5 (some hypotetical average move for some currency), then X>(3/5+1)/2 or in percents X>80%.

We need very biased stats to gain profit with this system (or very volatile currency, where N is big).

-- edit --
Little correction in formulas, it was late night and some errors slipped through. We should not use spread in recursive formulas, but after them. Thereby first result would be
P=-S (profit equals negative spread value)
and second formula (without spread)
P=X*(P+N)-Y*N => P=N*(2*X-1)/(1-X),
P>S => N(2X-1)/(1-X)>S => X>(N+S)/(2N+S)
for N=5, S=3 we get X>8/13 or X>61.5%.

Much better - to get profit iin this situation we need statistical edge about 62%
N00B ATM
  • #20
  • Jul 11, 2007 8:08pm Jul 11, 2007 8:08pm
My apologies, Inflection. I inadvertently hijacked your thread.

I may just move this whole coin tossing thing to my own thread. Or maybe I should just attach a lightening rod to my head.
You are in a maze of twisty little passages, all alike.
0 traders viewing now