With that out of the way lets get started.
According to Thomas Bulkowski's research in his book the Encylopedia of chart patterns, he goes over one pattern... high tight flags.
This pattern he identifies with "break even failure rates" which basically give you the percentage chance that a pattern fails to reach a threshold. As such, those remaining reach at least that threshold.
For example, if 33% fail to reach 35% that means 66% do. If 67% fail to reach 75%, that mean 34% do. He also lists 58% rise over 45%. This seems consistent with my understanding as 46% fail to reach 50, meaning 54% do. But worst case scenario, with an 8% stop, and 25% trailing stop that replaces that stop, you have a profitable system as those that rise over 45% even after a 25% pullback will still produce gains.
It's important to realize that you cannot simply subtract 45% by 25% to produce the estimated gain after a 45% rise followed by a 25% pullback. If for example a stock went from 100 to 145 and then pulled back 25% it would not pullback to 125, but instead .75 of 145 or $108.75. It's important this distinction gets made to avoid overestimating returns.
I propose a system with a 25% trailing stop. The actual determiner of whether or not the "trend changes" is a 20% decline, not 25% so some of the numbers may be slightly different. Just keep in mind that with a 20% trailing stop the numbers are the same, but the losses are less and the gains are more. The derived results and probability of them occurring are estimated as follows
0% lose 25%, 2% lose 21.3% 4% lose 17.5% 4% lose 13.8% 7% lose 10%. 6% lose 6.3% 10% loses 2.5%. Finally we "enter the black". 13% gain 1% 21% gain 13% and 33% gain 31%.
Note that due to possibility of gap down, that will only hurt the performance and gap-ups will never help since it is a stop and can only stop out LOWER than intended. Nevertheless if this data was accurate, what kind of gains are you looking at?
You will be astounded.
You are looking at if you allocate 100% of your capital towards every opportunity and make 12 of these in a year at doubling your money on average. The difference between the 25% trailing numbers and 20% trailing with the same numbers is in a year tripling your money instead. The actual data is based upon making such a move PRIOR to a 20% decline from the high. As such, I believe it's entirely possible that you return more with the 25% trailing stop since the actual numbers that reach certain thresholds improve. If you are looking at the percentage that reach at least 75% and you allow for 24.99% pullbacks rather than 20%, some more will reach that threshold and the results will be improved.
But we can improve this system's expectations. Afterall, if we assume those that hit 75% simply retrace and we have no information on how far beyond 75% those make. The wording of the table "over 75%" is confusing because it says 67% fail to reach 75%. Therefore the remaining 34% do reach 75%. What does it mean that 100% fail to reach "over 75%"? Perhaps they fail somewhere between 75% and 100%? Perhaps no threshold beyond 75% was given? I don't really know.
If I am interpreting this correctly, at a minimum we know that 34% hit 75%. But looking at these numbers, one thing is apparent. If we wait for those to hit 75% and pullback 25% or to 31.25% from our entry, we are making a mistake. As long as we are calculating zero to go beyond 75% and 34% to hit 75% exactly, we would be better off taking profits rather than waiting for these to pull back. The same is actually true for ANY trade that goes above 31.25%.
Now lets enhance. Depending on your interpretation of the strategy, you could take profits when 30% is hit or 35%. I am going to instead go straight to the book where it says 58% rise over 45%. As such 58% of the trades we will take profit immediately at 45%. But using this data will screw things up a bit. Afterall we already have done all the calculations. Instead what if we just took those that rise over 50% and take profits at 50% exactly. 53.75% of the time we hit that target.
0% lose 25%, 2% lose 21.3% 4% lose 17.5% 4% lose 13.8% 7% lose 10%. 6% lose 6.3% 10% loses 2.5%. Finally we "enter the black". 13% gain 1% 21% gain 50% and 33% gain 50%. Do you see how "enhanced" this strategy is? 514% over 12 trades. But lets be more exact. The exact number of trades in a year with no overlap is limited by the number of days per trade. The "days to ultimate high" is listed as 39. I do not know if this is "trading days" or actual days. But if we assume that it's "trading days" there are then around 255 trading days a year divided by 39=6.5 as opposed to 365/39=9. So we actually were high in our estimate of 12 trades a year. If we instead use just 6 and round down, we get 148% per year with 100% invested.
Now lets see if we can enhance further. What if instead we took profit at 35%? We then have 126.7% per year. What about 30%? 128.9% per year. 25%? 117% per year. 20% 102% per year. It's actually possible we might be enhancing with any one of these strategies simply because we haven't looked at the optimal amount to risk, and we've "capped" it at 100%. Nevertheless, it certainly seems that this does not enhance returns yet. What if we left the trailing stop unless we hit 75%? 137% per year. So it would seem taking profits at 50% is probably best, and leaving the 25% trailing stop in.
But wait, we already recognized that only a 20% trailing stop needs to exist for the probabilities to remain the same. Now with a 75% return and a 20% pullback, we have a 40% gain. Could this skew our results to where the 50% target is not best? Lets find out.
First, the trailing stop strategy on it's own over only 6 trades still produces 103.7% return.
Taking profits at 75% only and letting the trailing stop works of course boosts it. but to 193.2%
Taking profits at 50% takes the yearly annualized profits to 187.7%
35% target takes profits to 151.86%
30% target takes it to 146.7%
25% takes it to 128.85%
20% takes it to 108.24%
15% takes it to 80.8%
10% takes it to 52.95%
5% takes it to 25%
Now this is perhaps not a fair assessment of the data for a variety of reasons. first of all, 6 trades total to ultimate high (before a 20% pullback) since we don't know how long the pullback from ultimate high takes, there may only be 5 trades a day. However if we assign profit targets, the closer the target, the quicker profits are taken, and of course, the more trades in a year we are able to do. Additionally, with the 5% target, according to the data, we never take a loss. Even though that is not realistic and only was the case during a great testing period, if that were true, you could safely apply an infinite amount of leverage in theory and as such, the actual strategy would be much, much safer at 100% of capital. Proportionally to risk it would offer a much better equivilent of reward to risk.
There are still ways to improve the strategy, but it becomes increasingly difficult to estimate. In other words, once a stock goes above say 35% for example, you would use a discretionary stop or tighten up the trailing stop even. Or perhaps you would set a hard stop at 30% at that time, and manually move the stop up over time. The goal would be to let it run but not let the large gain retrace a full 20%. In theory you could try to time the top at THIS point since you have already done the work of letting the winner run to reach the threshold where it's still VERY profitable annually to just let it stop out at 30%. But if you are able to let the winners really run and time the top AFTER the 30% threshold, you may be able to drastically improve your results. The idea is that the returns will be likely somewhere between 30% and 75% at that point, rather than just a flat 30%.
For example, the stocks that peak out at 35% you might only gain 30%. The stocks that peak out at 50% you might average 40%. The stocks that hit 75%, you might average 50%. If this is the case, you hit 203.16% on the year over only 6 trades on average with 100% invested per trade. With just 59% of the capital at risk, you still will expect to yield 100% on this strategy.
Typically I like determining the "full kelly" and comparing all strategies to it. In other words, One strategy the ideal betting amount may be 20%, in another the ideal may be 40%. Rather than comparing the results at 20% of capital at risk to 20% I would compare 1 full kelly to 1 full kelly. That way you are applying the "maximally aggressive risk strategy to the maximally aggressive risk strategy then you can adjust afterwards if you like. However, when the kelly is beyond 100% I don't like adjusting it since things get out of wack and since the formula doesn't account for leverage costs and risk of an intratrade downward that could result in margin call. In no circumstance would I ever suggest risking more than 100%.
As such, the real test of any of these strategies would instead be to consider an option strategy that follows along the underlying, and can be compared to similar strategies based upon an options calculator calculation estimation, but that is another story.
-----------------------
disclaimer:To re-emphasize the disclaimer on the bottom of the blog on every post... All math done is believed to be accurate, but could be way off. Much of the data was compiled by using the data contained in Encyclopedia of chart patterns and were then applied as only my understanding of how the trades would play out. The data was only applied to a spreadsheet which is only my understanding of what I believe to be accurate math. The actual expected returns are only based upon long term growth rate as the number of total time periods in which the given amount of trades make approach infinity, and as such the calculations should not be taken as actual expectations as no individual has the lifespan to live out "infinite trades". Instead, it is used only as a comparative measurement to ATTEMPT to assess the more profitable strategy with regards to risk management, and not actual suggested capital at risk per trade. The data is also based upon the past and it has not been determined if going forward, the data was still as profitable. This and many other risks must be considered and this is only a trading journal of realizations that I have made from which to improve my own financial understanding. These trades are not a recommendation to buy or sell any security or consider using any such strategy on your own.
No comments:
Post a Comment