I have a system in mind: - enter a trade at set times - every 20 minutes - fixed SL and TP where TP = 2 x SL - SL and TP distance from the entry price decided by the market range over 20 minutes (100-period moving average) so that a trade does not last longer than 20 minutes, statistically - long/short - decided by a coin flip - position size - increased when daily PnL reaches certain value Do you think this system could be profitable?

But the market can move violently in one direction or the other over any given 20 minute period, or it can move not at all for hours. It can not move for hours, and it can then move crazy for an hour, then it can crawl to a halt for another hour. The key would be placing correct entries. But you stipulated that long or short is a coin flip. So 50/50, plus the 20-min time constraint. Just set up a backtest and see how well it works in whatever instrument you wish to trade.

OK. Let's leave trading aside. Imagine a game with following rules: - we flip a coin: heads - you win, tails - you lose - you decide on the size of a bet - when you lose, you lose your bet, but when you win - you get 2 x your bet Would you play this game? I claim that a player can win this game by adjusting his bet size, depending on the series of winners and losers. That is the secret to this game, and to trading/casino etc.

If it is just a game with no real money on the line, sure, I'd play it. But trading is not a game, it is a business. I understand the logic behind your query, and the simple approach you are taking to this idea. Adjusting your bet (increasing/decreasing position size) will do nothing but amplify/lessen your gains and losses. It will not in anyway affect the outcome of the system.

@Overnight is 100% correct, this strategy is a no go. Lets say there is a 50/50 chance in the next minute the stock will go up. How about the next minute? Well it's again a 50/50 chance and how about the 3rd minute? another 50/50? This is what is known as a martingale process. So to go up 10 minutes in a row you end up having a very small chance (.5^10) Looking at the big picture. A stock is trading at $100 it has a 50% chance it goes up and a 50% chance it goes down. So you put your SL at $90 and your TP at $120. You are risking 1 to make 2 (TP = 2x SL). The odds are 50/50 it goes up or down, NOT that it goes up X $ or down X $. Using are first example, the stock will go to $90 about 2x more often then it will go to $120. Meaning you have an expected value of a big donut.

The trick to trading is to find the points that are not random. Perhaps 75% of moves are random, knowing how to identify the 25% that are not random....thats the game

What appear to be random over a long period may not be so random in the shorter period. There are certain period what head or tail will appear consecutively in coin flipping. Based what I have observed on stock direction, trends with longer life span exhibit low price variability. The inverse is also true. This is the frequency vs % daily change bar chart for DJI. 2008 is included to show worse case example.