wow - how to get trapped - Lol. Everyone else is already long before the break.

Really? Thanks, David

I like to use 3 moving averages and seeing how the close relates to them, under and over etc. Those and the activity of volume over the same days as the averages. Thanks, David

interesting, but there is a flaw, imho. Your position sizing depends on the previous trades' success or failure. You state "what are the chances of a losing trade after three losses" ? Well, these are the same (75 %) as each trade is independent from the previous. You can perfectly loose ten times in a row, and still end up with a 25 % winratio in the long run. So, that approach is much like playing in the casino "red" or "black" and martingaling. Beware of "the house".

kind regards, Marc

Marc Verleysen, nice point to bring up. So how would you describe to someone the difference between say gambler's fallacy and mean reversion? Also, if majority of mkt participants believed in the same way - erroneously or not - that previous events affected the outcome of the upcoming event, then could i.i.d still be assumed?

Hi Phillip, first of all, I am not a scolar or professor, so do not expect any scientific reply.

I have seen too many accidents happen with Martingaling. It often works, but the one time it doesn't, it bankrupts your account. Overall, I remind people that martingale or reverse to the mean as a strategy, of Maynard Keynes who said "The market can remain irrational longer than you can stay solvent". I am not here to say that variable position sizing is an absolute "no go". I am just here to warn of this use based on a back test.

If you assume that past events/trades have no statistical significance in terms of predicting future outcomes then you are implying that backtests don't make sense at all. Is this really your opinion?

Marko, I think you misinterpreted what I wrote. Please reread my comment on raising the position size after a series of misses. Backtesting a strategy is a "conditio sine qua non" before engaging in live trading. However, "double when in trouble", is the fastest path to financial disaster. Besides, it takes a very strong mind to play with a big position after having suffered a number of consecutive losses as one might start questioning the strategy.

IMHO, a strategy should work well with a "fixed position size". If you need to raise position size to make up for earlier losses, your trading signals need a review, not the size of the "gamble"

I've heard the anti martingaling point of view many times (and I even subscribe to this point of view for the most part)

However, just to play devil's advocate: Marc Verleysen let me ask you this

There are traders who trade the trend, there are traders who trade mean reversion, at the same time there are text book practitioners who claim the market is a random walk. no serial correlation, no impact of the past on the future? Applying a casino analogy to the stock markets is akin to saying that the markets are a random walk. (why trade them then?)

So who is correct?

Can't we just do a mean reversion test on profitability of a trading system to prove statistically whether the profits of the system mean revert (and if they do then over what time horizon)?

Also from a logical perspective instead of doubling down after a loss

why not just change the trading system to take trades only where there has been a loss?

The most import part of all is left out . See where the puts/calls ratio starts, They only start plotting on minute 15 after the open, trading before that is gambling, the starting position will tell you where the institutions have hedged thus giving you their overall direction of the day, next look to see what direction the P/C heads in and at what speed, this will tell you if the gap is going to be filled or not, if there is one. institutions trade from liquidity to liquidity , plot the POC on your daily, 4 hourly, hourly and 15min, this will tell you where they are going to, if the top POC gets rejected, the lower POC is the target, read Order Flow at each POC with volume to see if its being accepted or rejected.

Any backtest strategy figures sharing with us?