Whilst numbers are important, often the spur for testing is an idea conceived by pure reasoning or observation. After all, for testing to work well, first you want something to test. A person who is prejudiced against pyramiding/averaging due to flawed assumptions, or someone who does not really understand the logic behind it, will be at a serious disadvantage when it comes to deciding what setups and trading methods to test and then apply. Just look through this thread and other pyramiding discussions for examples - most posts are just based on old wives tales like "never average a loser" or psychological biases such as "average price" or open P&L. Once someone realises that trade size should be related to trade expectation and risk at the current market price - and that entry price, "average" price and all that malarkey has no bearing at all on current trade expectation at tht current price - then they are on the road to informed position sizing, instead of relying on half-baked trader lore and guesswork.
Chande's "Beyond technical analysis" has studied the above topic with equity curves and data for a comparison of 4 strategies: Constant 2 contracts, Half-on-loss, Double-or-half, and Double on loss. Perhaps defining what's the maximum risk in terms of % capital would be the most important issue above all, in order to avoid overtrading.
Thanks OddTrader. That Kaufman book looks very interesting (you can see the TOC on Amazon). It did not seem to attack position sizing as a separate topic. [And I saw something wrong in the "Basic Concepts" section where he implicitly assumes trading results will follow a Gaussian distribution and goes on to say if you have 400 trades the error in your results will be 5% (the sqrt N). Well, I've never had a Gaussian, please raise your hand if you have.]
Thank you sir. I have always respected your posts. (yes, I am still here reading every word and every post.) Michael B. P.S. average price is merely a float, stops and targets merely measurments of stage. Velocity determines exits.
Guppy suggested in his Better Trading about the above topic to analyse a trend into 4 stages: 1st- Young for always 3 positions, 2nd- Robust for adding 1 to 3 positions, 3rd- Mature for adding 1 or 2 positions, and 4th- Collapse can be sudden or gradual.
Roger that. Here's what I'm thinking about now: (0) You can think of adding/subtracting as separate entries. (1) Why would you want to add? Tentative answer (TA): if your original entry is a low probability one (e.g. top or bottom picking) AND you can identify a point of progress (profit) that IF reached, gives another entry. (1.1)If you can identify such a point, why not just wait until then in the first place? Or why not just enter full size to begin with, and get out if it doesn't reach that point? TA: If the low probability entry pays well when it works, pays well in the long run, but can subject you to drawdowns you'd rather not experience, adding on later can reduce the volatility of your equity curve, at the sacrafice of some profits. Not entering at all until the add point should also make money, just less of it. (2) Why would you want to take off (1/2, 1/3 etc.)? TA: If you can identify a point where the your move is likely to be over and take some profits there, you seal some decent profits if it does in fact retrace. (2.1) Why not just exit the whole position if you've found such a place? TA: If you have no means for re-entry, you allow for the chance of a further move. On the other hand, if you do have a means for re-entry, it's probably better to fully exit. (3)What if I can't identify such points or don't want to alter the volatility of my equity curve? TA: Then don't position-size.
Peter lets examine another thought and Thanks for your post, 2-TIER Primitive Example: Our hypothetical instrument opens at the price of 6 (one/tenth point increments) and our hypothetical instrument has a current "10 day ATR" (average true range) at 16 points. We are long 5 units from the open price of 6.00 with a stop @ 4 and a target @ 12 which represents a 1:3 risk/reward ratio (we closed yesterday neutral in the middle of the daily bar) The price goes up to 6.00 we do nothing. The price goes to 7.00 we "add to" 3 units (half of original size) So this gives us a stop@5.00 and a target@13.00. The ATR supports this so we have a trade that makes sense and is possible. Everything is moving fine and glad to have a spreadsheet to quickly figure this out as this baby moves fast. The price gives some distance from the stops and observing that the trend has given a nice big green bar. Hopefully, the trader is asking will it continue? How much do I want? This has advanced so quickly why not take it? Every trader has a different perspective about velocity, this is what makes the market work. I am just pointing out one scenario out of hundreds and the key is to advance your position to Lock and Load. So lets continue. The price is approaching 9.00 and the pyramid is looking like a good decision as the net position is 8 units and the average price is a blended entry of a Locked 6.40. math: (price of 6.00*5units)+(price of 7.00*3units)=51/(5units+3units)=an average price of 6.37 rounded to 6.40 At 9.00 our net profit without commish is 20.8, but if we just bought 5 units from the beginning at 6.00 without adding too (or pyramiding) we would have 15.0. But you say wait.... the risk! Ok, lets take it to tier three. The price is slowing at the price 9.00 your multiple targets and stops are intact. You know your average price is 6.40. Change you STOP ONLY on both trades to BE. Then wait for targets or continue to Load (or "add to") with a BE step after each tier. (even better lock an incremental profit with a BE plus some progressive profit with each tier) ADVANCED TRADERS Even on the way down in tier three from 9.00 you could load giving you an excuse to bring in your stop tighter with some BE plus profit. (if it was just a pullback then your winning both ways, as you really do not know in advance which way it is going.) I know, I know this is not conventional thinking...But pleeeease break out of the box! hope you are not confused. Michael B.