Candlestick probability
- Thread starter Amatrue
- Start date
You don't prove or disprove anything by showing anecdotal chart examples. You start by tabulating contingency tables and probabilities over large sets of examples .
At least Bulkowski did some of that for you. Start from there, not selectively choosing charts.
I was hoping the OP would have posted some chart examples but there was no charts posted unless I missed something.
Also, the last time someone posted a thread like this...he actually had statistical data with over 10 years of statistical data. Unfortunately, after all that hard work and data accumulation...his candlestick pattern identification was incorrect for many of the candlesticks....ouch
wrbtrader
I was hoping the OP would have posted some chart examples but there was no charts posted unless I missed something.
Also, the last time someone posted a thread like this...he actually had statistical data with over 10 years of statistical data. Unfortunately, after all that hard work and data accumulation...his candlestick pattern identification was incorrect for many of the candlesticks....ouch
wrbtrader
That 1 that went I presume up then gapped up higher = Bearish, just a buyers pause maybe, 50/50 crap shoot ofcourse.
Momentum, place you enter with x risk with okay odds of making X profit, repeat!!
but.... what if it does?
%%
I had to edit your word ''prediction '' out, Amatrue, AmTrend.
Mostly true. A green monthly candle, yearly up candle, in QQQ, SPY, tends to trend up again, bull markets. That's how investors tend to make millions in an uptrending bull market.Hint ;study a 200 +year chart of US stock market.[2] NO its not really a shift in sentiment, a 200 year uptrend is an uptrend+ that is why so many hedge funds lose money, blow up + underper form SPY. Wisdom is profitable to direct.......................................................................................................
Some things to consider:
- There has been an incredible amount of processing power thrown over the years at predicting the fixed time span price difference from a fixed time span lookback window (or simply the direction given a minimum level of volatility), because this is literally the easiest possible thing to program. As far as I can tell in my own testing way back when I was naive, this is basically as good as a random prediction without further conditioning on other variables.
- The market does not contain a whole lot of undiscovered inefficiencies just waiting there for not terribly original people to discover, especially without novel input data. Asking for a rationale on why something should pay you a return is wise if you want to avoid wasting time.
- You can certainly manage to predict that the probability of a green candle is higher than that for a red candle on 1s candles in an index future, and that the expected return is non-zero, but that's an entirely different thing from being tradable as slippage is much higher than returns on such trivial properties of price data on these short time spans.
Last edited:
you can research markov chains. To test will require a lot of candles