The Best Way To Choose Lot Size

Now, before I catch any flames for this, the title is intended more to promote discussion than my own methods. I will write about my ideas on the topic of position sizing, and will accept any criticism or alternative ideas from the many great minds on this forum (that's serious, not sarcastic).

Okay, to start off, yes I have read the thread on money management in the beginner section. Do I understand it completely? Probably not, but I have the jist (sp?) of it.

Now, as far as position sizing goes, I started off believing in the gambler's fallacy, which applied to forex, states that if I lose with one lot, next time I should place an order with two lots (assuming 1:1 R:R and 50% win rate). If I lose again, then I should place an order of three lots, and so on and so on, until I win, which should put me ahead (although realistically, factoring in the spread, you will still be in the red).

I seem to recall someone saying here on FF that you should not increase position size like this when you lose, but rather when you win. However, the outcome is exactly the same as the gambler's fallacy, only you are waiting for a loss rather than a win. With the gambler's fallacy, applied either to streaks of wins or losses, you will end up on the other side you started on (i.e. you will be black after ending a loss streak and red after ending a win streak). Again, factoring in the spread makes this a losing proposition either way.

This got me to thinking about the logic behind lot optimizing code and whatnot in certain EAs. I ran one simulation of an EA where I had a huge loss on a trade with a position of 40 lots, which caused the code to "optimize" my lots for the next trade at 0.1! Of course the next trade was a win of equal magnitude in pips, but I was still way down in terms of equity, thanks to the "optimized" position size.

Now I am still not sure what the answer to all of this is, but I believe it lies somewhere in increasing position with each rise in equity, but keeping it the same after losses. I know in extreme losing streaks this could cause you to go bust, I think that this is on the right track to the "true" optimized lot figure.

I am open to any and all suggestions,
andersenws

You're not far off of a successful money management strategy. Keep in mind, however, that it greatly depends on the rest of the figures, particularly the win/loss %. Take a look at fixed leverage position sizing, it might spark some ideas.

Twinchell, if it depends on win/loss%, then how do I determine that? (I know how to, my question is what time period) Do I look at win/loss for the past month? year? 3 years? My system is incredibly consistent, with high win/loss%, although the R:R is horrible (though not as bad as Firebird).

Thanks,
andersenws

Not sure if you've seen this but you could try checking out Ralph Vince's optimal F (try googling optimal F money management). This is an increase lot size after winning approach but shows how to easily calculate the 'optimal F' leverage factor for your system and it's drawdowns from a sequence of trade history. He claims that the optimal F will achieve geometric growth whilst not sending you broke. However whilst mathematically correct it has been criticised by practical investors for being unrealistic in regards to drawdowns. It is not uncommon for it to have 70% drawdowns on it's way to achieving geometric growth. With this strategy, it shows that all you need is a profitable system and with the use of the optimal F you can turn small amounts into very large amounts. The speed you will do this is determined by the quality of the underlying system. It's achilles heel for me is that it assumes that the largest drawdown provided in the trade sequence you provide it, is the largest it is ever likely to see and therefore uses this to ensure the strategy won't send you broke. However if you see a larger one - it probably will ! For many, they calculate optimal F and half it !

Good question. I'm afriad I can't really answer it. Maybe this will help a bit:

http://www.forexfactory.com/forexfor...ead.php?t=8344

I will tell you that 50% is very important statistically. Also consecutive losers can play a big role in your drawdown, but not really in your overall profitability; which is just as important.

In my opinion a martingale approach to position sizing is suicide. If you continue to increase your lots as you lose, it's going to look real nice, for awhile anyways. The key, in my opinion, is find a happy median between fixed lot, or fixed fraction, sizing, and martingale. Note that this is not good for every system, only some will qualify.

Just throwing my uneducated 2 pips in here:

I don't think win/loss ratio is that important compared to expectancy. Meaning you could have a scalping system that opens random positions with a take profit of 10 pips, and a stoploss of 2000. I'm sure the win/loss ratio would be sky high, but you'd probably still have a negative expectancy. This would be an "eat like a bird, sh*t like an elephant" type system. Even though you lose only say 1% of the time, over time you would average a LOSS of ~10 pips (.99 * 10 - .02 * 2000 = -10.1)

On the other hand you can have a system that's designed to catch major trend reversals, with a very tight stop loss of 10 pips, and a 2000 pip take profit. You would probably get stopped out for a minor loss many times in a row, but on the occasion that you don't get stopped out, you'll tear the market up.

Thanks a lot guys, especially twinchell. Now I have another question. Say I have an EA with code that automatically ups the position size as I go. What is a realistic limit to be placed on lot sizes. Sometimes towards the end of a long backtest, the EA will be placing orders of hundreds of lots and I'm not exactly sure this is realistic, given my hesitation to a trade that size and/or my broker's hesitation. I was capping it at 40, but I was unsure as to what a reasonable limit was.

Please give your input on this issue.

Thanks,
andersenws

This probably won't help your EA coding unless you want to chage it a lot, but the old-school way I was taught was a simple formula of what % of equity you want to risk:

Acct Balance= $5,000 x .05 (5%) = $250....that's what I should risk in the trade due to acct size. Then divide $250 by my potential stop loss, i.e. 25 = 10 lots (or $10/pip on GBP/USD pair)

By theory, if you're winning, you will expand the number of lots each time and build up the account, but never lose enough in one trade to cut too deeply into your start up capital or revenues earned (depending on how far along you are), just by making a bad trade.

Obviously, if you are losing, your balance will start to decrease, but it will keep you in the game a lot longer, with the best chance of recovery if your win/lose % is in a rut for a while.

Do I do this consistently? yes and no......I use mental stops as an intraday trader 'coz I've seen too many whipsaws take me out on a perfectly good and well planned trade that could of (or did) profit. I set large stops when I have to be away. (i'm not afraid of losing my internet connection......I'm a 5 min. drive from a completely different one at my house or office, and vice versa)

Sometimes, if I'm a little unsure of the trade, I reduce the lot size and keep my pip counts for lose or stops the same, then re-evaluate the account balance after the trade to see what my max lots could be for next time.

All I can say, is it works for me....even with some flexibilities I allow for, I'm consistent, and it helps that I win more than I lose and my R:R is not out of whack (ie hoping for 5 pips, but willing to lose 90!!!). My account balance is climbing nicely, and I don't feel like I'll be blowing my account anytime soon.....

Best wishes,
Jim

I wouldn't worry about a realistic limit. Your goal should be % return and maximum drawdown. If you're surverying hundreds and hundreds of trades, trading a couple hundreds or thousand lots in the end might not be realistic in real life, but it should be in a backtest. This might sound contradicting, but base your position sizing on a fixed fraction or a fixed leverage and its not.

Don't let big numbers make you think its unrealistic, just worry about percentage. I could say my system makes a 100 million dollars in 3 years starting from $10,000. Everyone will laugh. Tell the same people you average 30% a month for 3 years straight and they begin to see it is possible. Some at least; some will still laugh.

I absolutely agree with Twinchell - maximising your % return whilst minimising drawdowns is the key. Large lot sizes tend not to be a problem with brokers upto about $20M - then it's a nice problem to have. However there are things you could do - you can break up your trades across multiple brokers, run multiple models that have slightly different parameters and hence will have similar results but will trade at slightly different times...

There is another point worth mentioning. Systems that up the lot size on a profit and have a feature of having more winning trades than losing but also tend to lose more pips per trade when they lose than when they win, tend to not scale down their lot sizes quickly enough to be nicely balanced. What I'm saying is that it is better to have a drawdown made up of lots of smaller losses than one or two large ones, as the system isn't being given the opportunity to scale back in the same fashion it accumulated the profits on the way up. The consequence of is that it is possible for a trade sequence of +50,+70,+50,+70 (+240) followed by -120,-120 (-240) to leave your account balance -ve. Therefore the opposite will be true for systems that increase lotsize after a loss.

If you send me your backtested EA results, along with the maximum desired drawdown and I'll optimize the position sizing for you. Shoot me a PM if you want.

Thanks Loki, but I can optimize fine. This thread is more a question of theory (although I do have a specific EA I am foward testing now). I am just trying to figure out how to size my positions. I have even given thought to just having static sizes. This, I feel, is the most realistic situation. When I run a backtest, I want to simulate what it would have been like if I had been trading that EA. Now, if I was living off of those profits, I would have nothing to live off of if I kept reinvesting them. Therefore, static sizing makes sense because I can withdraw my profits at the end of the week/month/year and my system will not be harmed.

I know I keep going back to realism, which some said to ignore when backtesting. However, my purpose in backtesting is to see whether I could see some money, real paper money, in my hand by using this system. If I keep reinvesting, my account may grow, but my hand will still be empty.

Open to thoughts on this,
andersenws

I personally tend to agree with your static position sizing thought more than your equity curve crossover method. If you EA is of the scalping variety, the win/loss ratio is probably pretty much fixed. Therefore increasing your position size during a winning streak would be like doing the same on a coin flipping game. It doesn't matter if you've won 37 times in a row, the next flip still has a 50% chance of being heads, and 50% chance of being tails.

The only way I could see the equity curve crossover method working better is if your EA is some sort of trend following system and there are periods where it "syncs" up with the market well, and other times when the EA zigs when the market zags.

I personally prefer "fixed fraction" position sizing, but I think in the end you're the only one on this board who can determine what the best MM strategy is. It al depends on what your goals for this trading adventure are (i.e. how much you want to have a steady income, versus how much you want the account to grow rapidly).

If you don't think that you can risk 3 lots ($30 per pip) and lose five times in a row, then you shouldn't risk it.

The most important thing is that you set a goal where you risk what you can mentally handle.

Well, let me share some thoughts & ideas in this great thread...

I believe that ANY single trading idea must start with the minimum lot you are willing to trade...This is your standard lot size...For example, some traders use the 0.1 ($10,000) lot as their standard, because their risk rules can't allow them to use more...Others have larger account sizes & are more prone to risk as a percentage than others, so they use the 1 Std lot as their standard trading size, so it differs base on the account size...

So ANY trading idea must start with the minimum lot you are willing to trade...And I am talking about trend following systems here...As time goes on & the market moves, you get several outcomes about the probabilities that your trade is going to be a winner...Yet I don't consider the word "Winner", because a winner can turn into a loser in a matter of minutes, what I mean is that you enter a trade & expect a trend to develope in your favour, when you start seeing signs of the trend, you are getting more success probabilities...Then the trend starts revealing itself, making the probabilities even higher...At this point, when the trend reveals itself for the first time infront of everyone after being hidden for a week or two, you start adding to your position (Doubling actually)...

A good trend will rarely retrace to your breakeven point, not after it showed itself out loud to everyone...At the beginning, when it is still hiding, it might swing up & down, bringing your paper profit or loss up & down, but after the trend shows itself to everyone in the market & you double your position, it rarely retraces back to breakeven...

So, now you have double the position size just as the trend is clear infront of the whole world on the chart...Now you wait & watch...There are hundreds of traders all over the world, both retail & institutional who don't believe what their eyes see & fight the trend in it's early stages...Those guys get burned out in 80% of the times & get rewarded in the other 20%, because sometimes the start of the trend is not as strong as it should be....Early trend fighters are there & I see them everyday during market hours...Those guys are a mixture of people..Those who think that the market rallied too fast too far after a deep downtrend & should go down & must be shorted, they include previous shorts who were following the previous trend & made lots of money out of it, but they still can't change their point of view about the market from bearish to bullish, so they keep fighting the market to bring it down, shorting it more & more, and as they are already holding profitable short positions, they don't consider shorting more as averaging losers, they consider themselves shorting the rally, so they are comfortable about their actions as their account statements are still showing profits...

This in my opinion is the most dangerous part of the trend...This is where the most risk resides...I usually stand with double my normal position size at this point, untill the shorts & sellers get fried more & more as the market goes up & more up...At some point, the trend accumulates people around it & shorts stop fighting & start thinking about finding a way to escape the trap, at this point, I add once again, but only half what I already have...

Finally, after this stage ends, 70% of the traders in this specific market turns bullish & everyone is buying, shorts are covering their positions & turning long at the same time & the market is attracting participants from outside who come just to buy & the trend strengths further...As soon as I realize that we are entering this stage, I double what I already have...

Untill finally everyone is buying & nobody is left to buy, prices are through the roof & everyone wants your long position at any price, you will have to check contrarian indicators & get out as soon as you can & be flat...If the trend was short, I don't bet on a reversal, but if the trend was lengthy in time & many market participants are stuck on one side of the market, for sure I will bet on a quick & sharp reversal & turn bearish using the same position sizing model mentioned earlier...


So to make the long story short, here is what I believe about the correct position sizing technique...

Start with the minimum.. 1 Lot for example...

Double what you have... You have 2 lots now...

Add half what you have... You have 3 lots now...

Double what you have once again... You have 6 lots now...


Of course, if the trend is lengthy, you have further room to add more, but I will never add up more than one third of what I already have every time I add up...


Thanks,

Nader

hey guys, try out this position calculator based on the criteria discussed in most of the above threads....

http://www.goforex.net/mmc.htm