R. Tharp's Trading Methodology

Status
Not open for further replies.
Hi Robert

Would you mind explaining how you calculate your trade expectancy using the discretionary approach you describe? With a strictly systematic approach to entry and exit it can be historically calculated easily using testing software, but how do you find it when you use such subjective trade criteria as support and resistance levels? Are you using a set definition of S/R that can be calculated in testing software? Are you analysing your own past trade data using this methodology? (a very painful method for finding positive trade expectancies I would imagine)

I've never liked strictly systematic approaches and believe the best approach is a mix. However, I have never found an efficient way of calculating expectancy for discretionary "systems".


Kirk
 
to calculate expectancy take a sample of your trading.

Each trade figure what your R (risk) is per trade.


I tried to cut and paste a spreadsheet I use but it doesn't look anywhere near the same here.

What you need to do is calculate your R each trade. R=
(your entry price minus your stop) * by the # of shares you bought. This is when you first enter a trade.

Most traders risk around 1% which is what I recommend. I do use systems that risk a lot more but 1% is usually a good example.

If you Bought IBM at 100 had a stop of 95 and bought 100 shares. Your are risking 5 points and 5 multiplied by 100 is $500. If you are stopped out you lost 1 R. If you fail to execute your stop or are gapped on at 90 you lost 2R or twice what you originally risked. Now since you are risking 5 points, everytime the stock makes another 5 point you make another multiple of R. so if you bought it at 100 and it's a week later at 130. You just made 6 R.

Just divide the amount won by the amount risked to figure it out. By the way most trades aren't even #'s of 1R or 2R it's more like 1.34R or 5.78R as stops are raised and slippage can happen. Don't round off the #'s. It is a lot more accurate to not round them off.

After about 40 trades add up all of the R multiples. Divide by the # of trades. This is your expectancy. It is an average of your profit you will make over the long term with your system. The more data you use the more accurate this is.
I tend to recommend first using very mechanical systems. With time (usually years) a trader starts to get a feel for the market--a sixth sense. This is when it is ok to start to slightly deviate from system.



Rtharp
 
Robert

Thanks, I am aware of the procedure for calculating expectancy and you have now clarified that you base it on actual trades. My problem with this is that if you are looking at potential trading strategies, there is no way to get the expectancy until after you have traded it for awhile (40 trades in your example which would be only barely statistically acceptable). With a striclty mechanical system, you can just program the system in to TradeStation or something similar and it will calculate it for you. I suppose you could paper trade your discretionary methods for awhile, but of course this requires large amounts of time to test a single strategy for expectancy and comes with all of the disadvantages of paper trading. I was just hoping there might be a better way. Thanks for your input.

Kirk
 
I should mention something Wareagle. I don't try to find new ground when trading. I find traders who are the top in the field and model their strategies. In fact because I understand expectancy once I understand the basics of the system they use I can out produce them in returns with less drawdowns.

Most business fail when they first start up, but 80% of franchises do well. It's because someone found a system that worked and taught others the system.
Anything I do has already been done by someone else. The system worked. I may tweak a few things to make it my own but nothing extreme.

Robert Tharp
http://communities.msn.com/rtharpland

 
Robert,

How are you able to talk to the "best" traders/fund managers? And how is it they are willing to share everything with you? Is this through your father's relationships with them? Certainly an advantage the rest of us will never have. Sounds like the opportunity of a lifetime, I hope you take full advantage of it.

Kirk
 
Well yes and no. I know a few traders who keep their addresses very private.


The majority of traders I know that are really good give seminars, write books, have a website. I just asked what they did. Spending awhile on the floor of the CBOT helped open a few doors. It becomes obvious which traders are doing well and who isn't after a year. (the traders that are still around)

so I have an edge but if you want the same it is possible. It's one of the reasons I'm doing my interviews on the website.

rtharp
 
You say there are lot of good Traders who have websites and will share their ideas and strategies. Which ones would you suggest that a intermediate daytrader model?

Trapper
 
Hello Robert,
I follow your posts with very much interest. I have also read Van Tharps book and I can nothing but agree with many of the statements and ideas in this book.

However, with regard to risk and expectancy in trading, I'd like to propose adding another dimension. I don't want to make things more complicated as they already are but IMHO risk is something different than just the amount of USD you may loose on any given trade if it hit's your stop loss.

Let's assume, you trade a certain system, which provides you with defined entries and exits. To keep it simple, let's say you trade Stochastics signals intraday.Whenver STO readings have been below 20, turn back to the upside and cross 20 line, that's your long entry signal.

An entry signal is given and you enter the trade, and after getting filled, you enter your stop loss. In order to keep your risk low, you take 1/4 point stop-loss on 1000 share lot. So your risk seems to be clearly defined - it's 250 bucks.

However, in my opinion, the real risk is based on the probability, that your stop-loss will be hit or not.

If you see i.e. on your charts, that in 7 out of 10 similar STO entry signals in the past, the stock dropped another 1 point before finally confirming the signal and going up a full point or more above the original entry price at the time the signal was given, than the probability of loosing the 1/4 point would be relatively high.
If you would use, based on this statistical probability,say a 1 1/4 point stop loss, the stock would tank but has a high probability of coming back without having hit your
stop-loss.

In other words - the probability that the stop-loss get's hit defines the real risk, not the amount of $ you are willing to risk. Because you can loose 4 times ( or more )the 1/4 point before the signals turns out to be valid.

Now you may say - well, I'am smart enough to recognize this STO pattern and I will wait whether the signal turns out to be valid. But in this case, you "risk" to miss the optimal entry point because this time the stock could go straight up without tank a point before validating the signal. If you're entry is 1/4 or 1/2 point above the optimal entry, your profit expectancy on the move up is lowered by 25% to 50%.

Wouldn't it be smarter to adjust the size of your position according to the probability that the trade will be profitable instead of using just a fixed stop-loss of say 1/4 point or 1/2 point on a trade ?

If you would take 200 shares instead of 1000, you could give the entry signal the necessary room to validate itself without risking a higher amount of money than with your initial system.
Even though the total profit may be smaller than with 1K share lot, the probability of reaching this profit would be much higher.

You could apply this also to other trading-systems, i.e. if you trade on S/R breakouts. If the stock failed to manage a real breakout 5 out of 10 times, than the risk is 50% that your stop loss gets hit when it's to tight.

So when determining a stop-loss level, you may have to take the probability based on statistics that it get's hit into your risk / reward considerations, as well as the appropriate trade-size according to this probability.

Looking forward to your thoughts with very much interest.





 
Privateer

Along the lines of your post, there is a concept that addresses setting stop losses called "Maximum Adverse Excursion". In a nutshell, it is an analysis of a system's WINNING trades and how much the market had to move against you before they became winners. You would then throw out the top 20% or so largest adverse moves and set your stop to encompass the remaining 80% of the moves, assuming the risk is still acceptable to you.

The theory behind it is that losers will be losers no matter what, but you want to hold on to potential winners as long as is reasonably possible for them to become winners. From this new stop level you adjust your position sizing to fit your risk parameters.

As with any method there are arguments for and against it. An argument against it is that you might eliminate some of your largest winners, but you might just as easily eliminate enough larger losses to make up for those. Anyway, there is a book on the topic called, oddly enough, "Maximum Adverse Excursion" by John Sweeny, although I have not read it. I learned about the concept in several other books/magazine articles, but I can't remember which. Perhaps someone else here (Robert or some of his traders, maybe) knows more about it or has used the concept successfully or unsuccessfully.


Kirk
 
Status
Not open for further replies.
Back
Top