Not having read everything in this thread, I believe that if you posses a truly proven strategy, you probably will have put your size till the point where your strategy starts to fail to further scale up. Applying Martingale to it simply wouldn't make sense. If you didn't apply maximum size, you either didn't have the money to put up or you had doubts about your 'proven' bit.
40yotrader is the only person on these boards to use a martingale strategy that i know of. Do a search for further reference. I think you would need to be trading 20+ instuments with 5% of equity allocated to each to survive the drawdowns and make it work though. Runningbear
using the martingale system (or any other betting system) will not change mathematical expectation. Will it will do is modify the profit/loss curve (eg. some systems will give you a higher probability of winning, and a lower probability of losing, but the avg loss will outweigh the avg win so expectation won't change).
Speaking of bet size, how many traders code pyramiding into their strategy and do they find it makes them that much more profitable? Its like the old turtle strategy going around. The only way some of those guys code have got big numbers every year was by pryamiding their positions up. Otherwise the system is pretty mediocre.
Almost all TA methods are unable to avoid long string of losses, and hence, definitely not suitable to apply Martingale. So the question is which methods can avoid long string of losses?
1. 30-35 trades a year tested 5 years, approx. 180 trades total. 2. has seen 5 losses in a row 2 times - 5 losses in a row, 60% to the power 5 is approx. 7.8% - 7.8% of 180 trades = 14 trades About the right size for the 2 times of 5 losses in a row! 3. to see a string of 9 losses in a row - 60% to the power 9, is approx 1% - we need 1% >= 9 trades. - So in 900 trades, there will be at least 1 such event. For lower count consecutive losses, like 6 times in a row, that will happen much sooner.
It is very difficlut to discuss these systems that were developed for gaming to trading situations because in gamming the payout is always known. In trading, the payout is unknown unless you always take profit targets and stop losses in a completely static way (say you always take +4 ES PT and always take -1.5 ES SL or any static variation thereof) probably a losing proposition in "the long run". Martingale Betting Systems (MBS) are also known as negative progression betting. Positive progression (Parlay) is where units are doubled after wins and reduced to single units after loss. Flat progression is obvious. Negative and Positive Progression betting systems assume "games" that are not independent trials, like BlackJack where the game has memory. If a game has no memory (each trial is independent) then no amount of money management will help. It also assumes that the strategy is static as in the case of BJ (the cards in the deck don't shift based on all bets,) a situation that is definitely not true in trading, as anyone that has taken a system live can attest to: The "opponent" is always adjusting to "your" adjustments and markets amazingly count only those that enter them with real positions, i.e., increases or decreases open interest. Further, even in the cases where there is memory (like trading certain markets, some more than others,) these betting strategies only affect players or traders "in the long run." Many on ET are investors, and are not professional traders who have many many trades per year. Therefore, few will place enough bets in their lifetimes to satisfy statistical criteria for "the long run." In short, I think it is incredibly hard to know how much to bet in a given situation reliably and you can probably embed any question about trading systems in that question. nitro
Could some one list out the probability of string of loss to compare with the following. Your chances of being involved in an aircraft accident are about 1 in 11 million. On the other hand, your chances of being killed in an automobile accident are 1 in 5000.
1/(2^12) is about 1 in 8000, so if you have a 50/50 chance of being right/wrong, you woul dneed to choose the wrong turn about 12 times in a row if you doubled the risk every time. I think.