There are a couple of problems with increasing size during drawdown.
Firstly, what are the logical inputs to trade size? There are only three: your current risk capital; the trade win rate; and the payoff of the trade, per unit risked.
If you are concerned with keeping drawdown below a certain level, then the payoff of the trade is not that relevant either, because drawdowns happen when you have losing streaks, not when you have winning streaks, and payoff ratios don't make any difference when you are having X consecutive losing trades.
So, the main input to sizing during a drawdown is the win rate of the proposed trade, and your risk capital. That's it.
Thus, the only way that martingale can make sense is if having had a run of losing trades somehow either i) increases your risk capital (an obvious absurdity), or ii) increases your win rate.
Now, if you are trading some kind of mean reversion strategy, and your research shows that the more extended a market gets, the better the trade odds/win rate become on each attempted bottom-fishing expedition, then increasing size could make sense, up to a point. However, you always have the risk that it just crashes in your face, so there is an inherent limit to how much you can martingale. And this is on a dependent series of trades. If you are just talking about a common garden variety drawdown where the losing trades are independent, then the only reason to increase size during a drawdown is if you get a super high % win rate setup after a drawdown.
For example, if I've had 10 losing breakout plays in a row and I'm down 10%, then Barack Obama resigns or there is a huge earthquake in LA and I can hit the bid on the S&P, I would happily bet more than 1% on that trade. Whereas if I get an 11th breakout play, I am not going to increase my size just because the prior 10 lost money.
Now, there is a second reason not to increase size during drawdowns, which is that the occurrence of a drawdown indicates a non-zero probability that you are losing your edge. If you are down more than the norm, it is possible that (for example) your breakout strategy no longer works well in current market conditions (or maybe has been arbed out of existence for all-time, because it became too obvious to other market players). In this case, not only should you not increase size during a drawdown, you should REDUCE size, possibly to zero (i.e. abandon the strategy).
Think of John Henry's simple trend following - it is now mediocre as a strategy, whereas in the 80s it worked well.
Finally, consider the risk/reward equation for reducing vs increasing size during drawdowns. If everything is A-ok, then reducing size will just mean you recover a bit slower than normal - in the long-run you will still get to new equity highs. If something is wrong, you will be better protected. If you martingale than you will recover faster, but you will blow up if something is wrong. Martingaling has career risk whereas reducing size does not. That's an unfavourable risk/reward tradeoff for Martingale strategies - heads you make a bit more dough faster, tails you are unemployed.
So, a drawdown means you should reduce size, unless you get a tried and tested high % setup. Naive martingaling is generally a recipe for disaster.
Firstly, what are the logical inputs to trade size? There are only three: your current risk capital; the trade win rate; and the payoff of the trade, per unit risked.
If you are concerned with keeping drawdown below a certain level, then the payoff of the trade is not that relevant either, because drawdowns happen when you have losing streaks, not when you have winning streaks, and payoff ratios don't make any difference when you are having X consecutive losing trades.
So, the main input to sizing during a drawdown is the win rate of the proposed trade, and your risk capital. That's it.
Thus, the only way that martingale can make sense is if having had a run of losing trades somehow either i) increases your risk capital (an obvious absurdity), or ii) increases your win rate.
Now, if you are trading some kind of mean reversion strategy, and your research shows that the more extended a market gets, the better the trade odds/win rate become on each attempted bottom-fishing expedition, then increasing size could make sense, up to a point. However, you always have the risk that it just crashes in your face, so there is an inherent limit to how much you can martingale. And this is on a dependent series of trades. If you are just talking about a common garden variety drawdown where the losing trades are independent, then the only reason to increase size during a drawdown is if you get a super high % win rate setup after a drawdown.
For example, if I've had 10 losing breakout plays in a row and I'm down 10%, then Barack Obama resigns or there is a huge earthquake in LA and I can hit the bid on the S&P, I would happily bet more than 1% on that trade. Whereas if I get an 11th breakout play, I am not going to increase my size just because the prior 10 lost money.
Now, there is a second reason not to increase size during drawdowns, which is that the occurrence of a drawdown indicates a non-zero probability that you are losing your edge. If you are down more than the norm, it is possible that (for example) your breakout strategy no longer works well in current market conditions (or maybe has been arbed out of existence for all-time, because it became too obvious to other market players). In this case, not only should you not increase size during a drawdown, you should REDUCE size, possibly to zero (i.e. abandon the strategy).
Think of John Henry's simple trend following - it is now mediocre as a strategy, whereas in the 80s it worked well.
Finally, consider the risk/reward equation for reducing vs increasing size during drawdowns. If everything is A-ok, then reducing size will just mean you recover a bit slower than normal - in the long-run you will still get to new equity highs. If something is wrong, you will be better protected. If you martingale than you will recover faster, but you will blow up if something is wrong. Martingaling has career risk whereas reducing size does not. That's an unfavourable risk/reward tradeoff for Martingale strategies - heads you make a bit more dough faster, tails you are unemployed.
So, a drawdown means you should reduce size, unless you get a tried and tested high % setup. Naive martingaling is generally a recipe for disaster.
