Trading Analytics

Trading with an EDGE is part of what gives a trader a majority of their profitability, however, there are human factors of judgement that are also at play here. Cataloguing trades for review will enable the avid trader to analyze and optimize their trading. There is a mountain of data to comb for analysis and arduous tasks at hand for the analytical mind.

EDGE – Statistically significant correlation(s) that has the potential when utilized by a trader/system to assist them in yielding a higher-than-average return of the total market average.

One test I like to perform in order to determine if the outcome of my trades were by chance or by luck is the monte-carlo test. It is recommended, but not necessary, that you have greater than 10k data points (trades) spanning across several years of trading for analysis.

What I am testing and how I perform these tests:

The monte-carlo test I perform is designed as though a bot has copy traded my account all year but allocated its own copy trades by a percent of a divisor matched to either their account total value before a trade or just to a specific divisor (Basically, two tests were performed, one with dynamic allocation and one without). These bots were designed to trade my historical trades in a random order under 10k simulations so that a distribution could be plotted and analyzed. Here is an output of 726 of my trades randomized in a simulation for dynamic (by current account size) allocations:

Here are distributions of 10k simulations of my 726 trades:

What I did here:

  1. Exported real trade data from TraderVue
  2. Scrubbed/Cleaned and formatted data
  3. Ran simulated trades using a percentage-based approach:

A. Created percentage gain/loss of trade data set by dividing each trade return by 50k (to get a percentage of account made/lost per trade)

B. Randomized data set and applied it to percentage-based trading simulation

C. Reproduced simulation 10k times (randomly) and mapped distribution

These simulations give the ability of the interpreter to gauge whether the trades were significant with regard to the net outcome of those trades. Basically, was it by chance that I am profitable this year, or is there a high likelihood that my account will approach 0% in the future given the trades I have made this year? The monte-carlo test can also give you a realistic estimation of draw-down in the account (as you can see my simulation plotted above shows a very lengthy draw-down period).

Another way to look at this:

% Chance of loss per trade: (roughly 20% (0.2) for my current records) = D (How often a trade results in a loss)

Average Loss (2.9% of total account) can be converted to L = 100%/2.9%, which roughly equals 35. This number represents the number of successive average losses it would take for my account to reach 0%.

DL (0.2^35) should represent, roughly, the likelihood of my account reaching 0% given that my trades are consistent with those I have used to generate the aforementioned outputs. My probability is very small, < 1%. This can be interpreted as a low likelihood that my account will reach 0% if my trading remains consistent with my prior trading (or I could have 35 losses and go bust). This does not prove that I have an edge! This is not with full consideration of the massive outlier losses I have incurred, but does utilize those data points.

To decipher as to whether it is likely that I have an edge, I will use the following formula:

(U x G) – (D x L) = If positive, indicates that average profit is likely able to overcome average loss as long as new trades remain consistent with those analyzed.

% Chance of gain per trade: (80% for my current records) = U (How often a trade results in a gain)

Average gain (1.3% of total account) = G

My numbers:

(0.8*0.013) – (0.2*0.029) = 0.0046, This is my current profitability given my current trades. This number is an indication that my average win-rate together with average gain per trade is able to overcome my losses, if I can maintain or improve these averages, I should be profitable next year. I can also take this number and multiply it by my number of trades to get a future total max return on investment estimate (by %).

*This number should not be used to definitively determine whether you are profitable but can be a valuable tool to gauge current trading progression. What this number indicates is that the average profits and losses taken into consideration with the win/loss ratio are overall positive. There are many other variables to consider and there is still a probability that my current trading if consistent will not be profitable in the long run. There are some outliers in the data that now exist due to poor position management on my part (ie. I lost 1/4 of my account in two trades, this was clearly not due to my edge and my current risk structure now accounts for these). With this in mind, until I remove outliers in the current data set and have an appropriate number of trades, I will not be able to accurately judge my current edge and trading performance in the market using this performance valuation (there isn’t enough data yet). Although, this does give me a general estimation that can become a basis for comparison in the future.

To interpret the above image: For every $64 I gain, I will also likely lose $36, with an average projected net gain of $28. From that $28, I will only be able to take away roughly $11.20 (Taxes & Fees Hurt!). Multiply these numbers by 1k or 10k and you can see why trading is lucrative for some.

Leave a Reply


%d bloggers like this: