Quote:

Originally Posted by **TheRumpledOne** I believe you.
Let me help you. Look at the previous post. You'll see a chart with an explanation of what to do. It's really simple but it works more often than not. Do not let the simple instructions fool you. It's NOT a trick. It really works. If you do not believe me, then watch it on your own H1 chart so you can see for yourself. |

Could we first perhaps adress this issue of consecutive runs of "UNIQUE" green / red candles as you seam to be placing rather a lot of emphasis on this ?. Personally I'm of the opinion that this particular aspect of your analysis is fundementally flawed.

I really should have demonstrated this with price data, just to remove all argument, but as time is limited I'll use a coin as a proxy, but I assure you it works equaslly well on any timeseries. I intended to toss a fair coin (in this case a 1995 one Maltese Lire coin) 100 times, but it rolled away under the desk into a tangle of computers and wires, so the experiment was sadly abandoned after 60 tosses (so yes I guess that officially makes me a ****** !)

The results as they stand where:

HTTTTTTHHTHHTHHTTTTTTTTHHHTTHTHHHHTTHHTHHTTHTHTTTT HTTTHHTHTH

On the basis of simple probability theory we'd expect a 50/50 distribution of heads and tails (or red candles v green candles) So in our case we'd expect 30 heads, and 30 tails. Take it from me, there are 26 heads, and 34 tails, but feel free to check emm if you wish

The same theory tells us we should expect approximately 30 instances with a consecutive run length of one head occurring, 15 instances with a consequitive run length of 2 heads occurring, 7 instances with a consecutive run length of 3 heads occuring etc

Averaging the data from our fairly rough experimental evaluation resulted in the following:

Basic probability theory suggest that we should expect to observe:

A probability of 0.5 for consecutive run length of 1, our experimental data indicated a probability of 0.495

A probability of 0.25 for consecutive run lengths of 2, our experimental data indicated a probability of 0.255

A probability of 0.125 for consecutive run lengths of 3, our experimental data indicated a probablity of 0.13

A probability of 0.625 for consecutive runs length of 4, our experimental data indicated a probability of 0.83%

If we'd tossed twice as many coins, the errors (which are only of the magnitude of 0.5% or so as it stands) would be further reduced. I trust that even you would agree that this simple experiment verifies that conventional probability theory is reasonably acurate ?

Conversely, you're methodology of "I only count unique candles" kind of falls down. If you look at the sequence above, there are only 6 unique tails by your definition, and 7 unique heads (and you'll notice that 2 of those 7 occur at the first and last positions in the sequence which means that on the basis of probability one of the has a pretty good chance of becoming non unique by your definition if this sub sequence became part of a larger sequence).

So whilst commommn sense tells us that the chance of getting a run length of one head or tail, is 50% and simple experimental verification proves this to be the case, by your definition, run lengths of a single head or tail will appear only 23% of the time. Even funnier still is that by your definition, something that should happen less frequently (and is proved to do so by experimental verification) by your definition leads you to the false belif that it actually occurs far more frequently than it actually does

These voodo statistics in which a run length of 11 can occur, without a run length of 7,8,9 or 10 occurring are quite simply complete and utter nonsense. Whilst its possible to analyse data in the way that you have, how in gods name can you apply any sort of conventional statistical analysis on something that makes little sense ?

I dont really feel as if You cant win this argument. You present statistics that say for example a run length of 7 greens rarely occurs for a particular instrument in a particular timeframe, so if price action dictates, go short on the next red candle (or probably more accurately, a candle where price action in a faster timeframe appears as though it might close as a red candle.) Now there's nothing wrong with developing a feel and trading price action, and I appreciate that its practically impossible to specify mechanical rules in these cases with respect to entry triggers, stops, targets etc. I also conced that this aspect of your system is probably what frustrates new traders who are generally looking for simple guidllines, where in reality non can be given. It takes time staring at charts day in and day out to develop an appreciation of these concepts, you either see it, or you dont, and no end of statistics is going to help.

However prior to these "set ups" which hopefully you'll agree have been proved to be based on the most bizarre and totally unreliable voodo statistics, there are often quite clear well defined moves of hundreds of pips. So why not use your skills at reading price action, and take a chunk out a strong trend, rather than attempting to scalp a few pips ?. If you are genuinelly trying to help new trades, Id suggest that teaching them to catch falling knives, whilst good fun isnt in all probability the best approach to take.

I really would be genuinely interested in hearing your thoughts on the consecutive candle statistics issue, I take the point that theyre only there as a potential set up, not as a trigger, but why the completely bizarre implimentation ?

regards

zu