Probability and random walk

Let's says i have 10 years of daily prices on a stock ABC. I do some analysis and I realise that, for example, if the stock increases 5 days in a row (close > open), 75% of the time, the 6th day will be a decrease. From that :

• If we compare with throwing a coin, if i have 5 consecutive tails, the probability to have a 6th tail is still 1/2 right ? So knowing that i have 5 days of increasing price, i still have 1/2 chance that the price going up or down because price are random walk ?
• Is the bias to think that those 75% will continue in the future and then i can't just say i have a strategy with 0,75 hit ratio ? How can i logically prove that this strategy is pure luck ?
• Now imagine that i look closer and i see that on those 75% cases, there is several groups of 3 or 4 cases each where it happens for the same price, as if there were some "resistance". I don't really give any interest to technical analysis, however is it possible that the probability space changes and it's not 1/2 of chance going down when it "hits" this resistance price in the 75% case ?
• you can do those sort of things and maybe find SOME cases where your probabilties do hold in the future. The question is more A) can you find it on enough stocks and B) when you factor in transaction cost and losses on the one's you're wrong on, do you still profit ? – mark leeds Mar 23 at 17:57
• why we can't stay on one stock ? also, if i put some stop loss pretty close of my order price, do i "bias" the probability ? – TmSmth Mar 23 at 19:27
• Hi: you can do it on one stock but that's a LOT OF RISK. Yes, you bias the probability with a stop loss but the chances of your probability estimate being all that good is not that great anyway so you're adding bias to an already volatile-questionable estimate. This is because, with these types of things, the past is not necessarily reflective of the future. There's no one behind the scenes making the stock price be up on the 8th day after 7 down days or vice versa. It's all kind of random. – mark leeds Mar 24 at 3:02