Thursday, September 17, 2009

"DayTrade"

8 days, 47 trades & +11.17R after my MAE- assisted light bulb moment, I thought I'd quickly share a few statistics and the possible implications these may have;

* 27 winners (35.91R), 20 losers (24.75R)
* The above means Avg winner= 1.33R, Avg loser= 1.2375R
* 57% (0.574468) win rate
* 0.2374466 expectancy.

First impressions are good. The only number that really matters to me is the expectancy...considering Roulette gives the house a 5.26% edge (0.0526 expectancy), I'll definitely settle for the above!

However, the other values can cause problems of adherence to/tolerance of the edge. One potential problem that was obvious to me was the effect of any decent drawdown. If the system is making a little more on it's winners than it's losing on the losers, then, (assuming a random distribution of wins and losses) would it be more likely to chop it's way up the Y axis, opening up the possibility for a pretty severe drawdown relative to the slow but consistent collection of R's??

As it goes, that question got answered on Tuesday 13th September :)- 6 wins, 10 losses (!), 6.23R down. 62.7% of the prior 5 trading days gain...(9.93R)

If that day's excluded, we have the following stats;

* 21 wins (29.92R), 10 losers (12.53R)
* Avg Win= 1.4248R, Avg loss= 1.253R
* 68% (0.6774193)win rate
* 0.5609936 Expectancy.

There's two main choices that I see: choose to ride out those ugly days or try to limit them. I've noticed that the other days huddle quite nicely around that 0.56R expectancy while that Tuesday is sitting all on it's own at the edge of the bell curve (-0.389R)

So I'm thinking, what if I apply the same concept of MAE to the WHOLE trading day, effectively treating it as a "trade"? Maybe below -XR my odds of making money are reduced enough to warrant cutting the "DayTrade" loose?

I'll continue to trade/collect data and see what picture it paints me...

Tuesday, September 8, 2009

MAE & Me- Using Data To Hone An Edge.



S0 after much struggle, it seems there could be light at the end of this particularly dark tunnel.

I've been trying to do something that I believe Know I can do. I just need to figure out the, umm, technicalities (ie how to do it...lol).

The chart shown demonstrates what I've always thought I could do but never had the conviction (or is that the data...?) to do. Even with a Sim account. This experiment came about by my study of the last 100 or so trades.

It quickly became obvious that my entries were so accurate, my Maximum Adverse Exposure, MAE, so small- 90+ % of winning trades never going into the red before showing a profit- that I had an untapped resource which could hone, if not stand alone as, my edge. What I was looking for was hidden in randomness- stops that were too wide to harness the strength I have at pin pointing short market sprints from fairly accurate entries...

Another obstacle is the tendency to watch dollar amounts rather than return on risk. If further along the trading path, you might watch return on risk but artificially cap it rather take profits at logical points. I suffered from this in particular and am still working on it. Eg I'd take half off at a predetermined R/dollar amount because I "couldn't risk giving back so much"- whatever that means!

The above problems of "giving room" with stops for such trades, watching where you are in a trade rather than where the market is in a move, result in no edge (gambling!). Fixing the risk will also leave less to chance (but will aggrevate any tendencies to watch dollar amount rather than the chart)