Martingale - How it can work

[Sigma-D]

Forex is one of the few areas where one of the barriers to success with Martingale are immediately removed: You can trade for all intents and purposes, unlimited size - there is no table or house limit.

The other negative factor cited for Martingale as making it a losing strategy is that you'd need unlimited capital. Given that the run of losing trades can be significantly reduced by only trading with obvious trends and for targets which are commensurate with trading timeframe and the facility to start small, it is quite feasible to successfully employ a Martingale strategy.

An example: 10 pip stop and 10 pip target. Your spread automatically puts you at a disadvantage, but trading with the trend and for a very modest win target negates that 'house' edge.

The 10 is not significant. Any number within the probabilities of a normal daily range would do, but there's no point going large as it's not the pips won that provide the profit but the target being hit - a winning trade, regardless of number of pips. Too small and your stop will get hit through random noise.

A 10-15 pip stop & target on a trending pair should yield a profitable Martingale over time.

Final issue is the consecutive losing trades, even with the above criteria in place on current market volatility (low) you could statistically expect between 6 and 9 consecutive losing trades. Which means you could be looking at trading X1024 your initial stake. But this too is manageable with mini and micro lots offered by most brokers.

The automatic response to Martingale as a trading strategy rarely considers the forex market specifically or the facilities for small capital outlay offered by most brokers these days. It's a case of the technology and facilities and even the market moving on but people still thinking of the standard responses to Martingale which were made in relation to theoretical casino and margin sensitive futures markets.

 

[Sigma-D]

Kelly and Martingale

Given the fixed reward/risk suggested above, the Kelly formula is highly sensitive to win/loss ratio. A 50/50 yields a fractional bet size of 0 which is quite sensible in those circumstances. Without altering the reward/risk, the only way t make this implementation of Martingale profitable is to ensure a higher win rate than loss rate.

A theoretical run of 100,067 trades taking spread into account using a very basic trend methodology provided a Kelly fraction of 0.09 (9%). Given the potential for a X1024 initial position size using this methodology, an initial risk of 0.009% is suggested.

Too small to be worth the trouble?

Depends how many trades you take in the trading day. The in sample analysis provided an average of 23 trades per day which yields an approximate return of 0.2%. Compounded over 200 trading days results in an annual return of around 49%.

I conducted the same analysis on out of sample data and obtained a statistically significant confirmation of the initial data.

Martingale can work.

 

[timsk]

Hi Sigma-D,
Interesting idea for a thread.

Speaking as a member of staff, I must make clear that the ‘official’ view on Martingale is simple: don’t do it! Almost all traders who employ Martingale strategies blow up their accounts sooner or later, for the two main reasons you cite:
1. There’s a house limit which prevents a trader from betting in sufficient size to recover previous losses.
2. There’s a limit to a trader's capital.
Underpinning both flaws is Keynes’ famous quote: “markets can remain irrational longer than you can remain solvent”. I’ve fallen victim to this in the past and blown up an account following an unprecedented move which wasn’t supposed to happen.

Having said the above, my personal ‘unofficial’ view is that it’s healthy to question the golden rules of trading from time to time to see if they still hold water. If I’ve understood your post correctly, you accept the two basic faults with it, but you think you’ve found a way around them? Allow me to tease out a little more detail . . .

Forex is one of the few areas where one of the barriers to success with Martingale are immediately removed: You can trade for all intents and purposes, unlimited size - there is no table or house limit.

I don't trade Forex, but I understand and accept the principle.

The other negative factor cited for Martingale as making it a losing strategy is that you'd need unlimited capital. Given that the run of losing trades can be significantly reduced by only trading with obvious trends and for targets which are commensurate with trading timeframe and the facility to start small, it is quite feasible to successfully employ a Martingale strategy.

That's fine so long as you're good at finding and defining the trend. However, if your first trade in the sequence is just as the trend is reversing - you could get caught out. This will almost certainly happen sooner or later?

An example: 10 pip stop and 10 pip target. Your spread automatically puts you at a disadvantage, but trading with the trend and for a very modest win target negates that 'house' edge.

Please explain how or why a modest target negates the house edge?

The 10 is not significant. Any number within the probabilities of a normal daily range would do, but there's no point going large as it's not the pips won that provide the profit but the target being hit - a winning trade, regardless of number of pips. Too small and your stop will get hit through random noise.

As with the trend, the volatility of any instrument is a moveable feast. Sooner or later, the profit target and stop loss calculations will go awry and you'll get clobbered, no?

A 10-15 pip stop & target on a trending pair should yield a profitable Martingale over time.

As per my earlier comment, how does one define 'trend' and how can one be sure it's not about to reverse?

Final issue is the consecutive losing trades, even with the above criteria in place on current market volatility (low) you could statistically expect between 6 and 9 consecutive losing trades. Which means you could be looking at trading X1024 your initial stake. But this too is manageable with mini and micro lots offered by most brokers.

Perhaps the answer might be to set a maximum number of trades taken to recover any losses (a sort of long stop loss in effect) and be prepared to take a big hit every so often. I'm just making the figures up but, for example, a 25% drawdown on a rogue run of consecutive losing trades would be acceptable if the profits made exceed this figure.

The automatic response to Martingale as a trading strategy rarely considers the forex market specifically or the facilities for small capital outlay offered by most brokers these days. It's a case of the technology and facilities and even the market moving on but people still thinking of the standard responses to Martingale which were made in relation to theoretical casino and margin sensitive futures markets.

I hope others contribute to the discussion as your proposal is an interesting one, although I'm not (yet) convinced that it's actually workable.
Tim.

 

[Sigma-D]

That's fine so long as you're good at finding and defining the trend. However, if your first trade in the sequence is just as the trend is reversing - you could get caught out. This will almost certainly happen sooner or later? Of course. There is always going to be a probability of the first trade in the sequence being a loser, less likelihood of the 2nd, even less the 3rd and so on. As I sated, statistically in current market conditions, you could conceivably get to 9 consecutive losers, but the probability of that is the somewhere near the inverse of the number of atoms in the known universe.

Is defining and finding a trend really that difficult? For traders trading purely directionally that’s surely got to be a stock in trade? I think most manage it quite well as it’s pretty obvious. Where the majority come a cropper is deciding where and when to get in and how much room to give it. FWIW I use a moving average covering the last 12 hours. When the price approaches it and bounces off, that’s a good hint. The slope of that moving average is also a clue as to general direction. I have an even longer moving average which serves when the price breaks through the 12 hours one. I will only trade in the direction consistent with the relationship between the two moving averages. You sit out a few good reversals true, but you also sit out a lot more whipsaws.

Please explain how or why a modest target negates the house edge? The house edge is the spread. If you bet a 15 pip stop and a 15 pip target, even with a 1 pip spread you’re already closer to Armageddon than Nirvana right from the off. You’ve got two aspects to combat that edge; The modest target and the faultless assessment of likely price development. Price is more likely to move 10 points than 20; 20 than 50; 50 than 100. The more modest your target the more likely it will be hit. The downside is the more modest your target the more modest your stop too – it has always to be a 1:1 for a Martingale. Stop is a function of speed with which you get into the impulse, distance from rational s/r level and distance from likely regression level. All of which you control. Too much distance? Sit it out. You always choose which trades to take.

As with the trend, the volatility of any instrument is a moveable feast. Sooner or later, the profit target and stop loss calculations will go awry and you'll get clobbered, no? No. Stop loss and therefore profit targets are always going to be calculated the same way. Volatility changes, sure. But that’s either good or bad. No way of knowing what’s going to serve you and what’s going to kill you. You never do. The worst that happens it you get a loser. You expect losers. That’s the point. Double up the next trade.

As per my earlier comment, how does one define 'trend' and how can one be sure it's not about to reverse? Defined (sort of) trend definition and recognition above, but how do you know it’s not going to reverse? You don’t. But neither do you care. If it reverses you wait till it proves it’s really reversed and you trade from the other side.

Perhaps the answer might be to set a maximum number of trades taken to recover any losses (a sort of long stop loss in effect) and be prepared to take a big hit every so often. I'm just making the figures up but, for example, a 25% drawdown on a rogue run of consecutive losing trades would be acceptable if the profits made exceed this figure. No, no, no, no. A perfect way to screw up a Martingale is to start messing with it. Don’t move stops in-trade. Don’t make arbitrary decisions on when to cry uncle. You play each trade 1:1.You risk an initial amount small enough with respect to your total capital available that means you can continue playing well up to the most unlikely number of consecutive losers.

I hope others contribute to the discussion as your proposal is an interesting one, although I'm not (yet) convinced that it's actually workable. Neither am I. Just kicking it around.

 

[Solas0077]

I was never able to profit in forex with a 10-15 pips stop. It was hit in like a few seconds. It was like a gift to the broker.I found out by actually losing money that I could be profitable after 50-50 pips for stops. But then I focused on trading ETFs but I may come back to it. AS someone mentioned,%K is very sensitive in win rate and unfortunately it is hard to find strategies with stable win rate. If win rate drops to 40% for a week and then rises to 80% for another week the average assuming uniform trades is about 60% but the 40% period can cause to you some real damage.