Tuesday, April 16, 2013

Using Tradervue for analysis of autotrading The Lincoln Million

Although it has been a while since my last in-depth reviews about Tradervue, Greg Reinacker has continued to build it into one of the premier trading journals.  It is elegantly designed and for all that it's capable of doing, it remains surprisingly user friendly.

Here are a few highlights of the improvements I've personally found valuable over the past year:
For anyone who is serious in improving their trading performance, maintaining a trading journal is not an option, it's a must have.  And I believe Tradervue is one of the best out there.


Over the past several months, I've started to focus primarily on swing trading, which means collecting a meaningful population of trade data for analysis takes a lot longer (vs. daytrading).  And as I wrote in the prior post, I'm also focusing more on autotrading a large portion of my account.

Recently, I've gained enough data from autotrading The Lincoln Million (TLM) to perform an analysis.  TLM is a swing trading service where Doug has gained an impressive return on his own account, and one of a few services being autotraded in my account.

I've taken the The Lincoln List's Champ Camp trading course (worth it, recommended), and have seen Doug in action live within The Lincoln List and The Lincoln Million trading rooms (both daytrading and swing trading).  He has been very consistently profitable, but how would the autotrading results compare?


Here are the actual results of The Lincoln Million's autotrades from January 1, 2013 through April 11, 2013 generated from Tradervue.  Although I've been autotrading the service since November, my $'s allocation per trade were not quite consistent and locked in until the beginning of this year, so I excluded that group of trades.
High winning %  (82%)
Solid profit factor  (2.10)

Potential concerns:
Average winning trade smaller vs. losing trade ($89 vs. -$414)
Average hold time of losers much greater than winning trades (26 days vs. 5 days)

The report below displays a graphic summary of the Win/Loss ratio as well as the P&L and drawdown curve:
The report below represents another view of the how long winning trades are held vs. losing trades:
One interesting analysis method within Tradervue is the ability to analyze your stop locations.  I didn't expect the chart below to so clearly delineate a potentially more advantageous stop location:

The past 5 months represent my first experience into autotrading, so I really didn't know what to expect other than mostly hands off approach, like a mutual fund.  There are some other services I'm also autotrading which are based on writing options credit spreads and each service has their own pros/cons.  I'll likely write about those in the future once I have more data.

With regards to autotrading via The Lincoln Million, here are a few thoughts:
  • There have been a decent number of trades that The Lincoln Million alerted and got filled on his own account, but were not filled via autotrading.  As expected, nearly all the missed trades were winners
  • Primary reasons for no fills: short shares no longer available, signal to execution time lag (generally takes 3-10 minutes for autotrade order to arrive at my broker via Global Autotrading), and impact of orders on thinly traded micro and small cap stocks resulting in slippage
  • Unlike trading the very liquid ZB or ES (T-bonds or S&P e-mini futures) products where you can put on large size per trade ($1 million/trade is nothing), trading microcap stocks often have a much greater issues with liquidity.
  • Therefore, actual results from autotrading is likely <= 50% of posted results
  • However, if you are following the chat room alerts in real time and manually trading, I believe it's possible to achieve performance 75+% of posted results
  • Better yet, once you learn the setups, you can probably do very well trading your own stocks, which is what Doug ultimately wants you to accomplish
I admit, I've experienced frustrations watching unfilled trades become winners, getting bad slippage on fills, experiencing underwater trades that are held multiple times longer than winning trades, and simply seeing average winners be smaller than average losers.  My initial expectations on what the returns from autotrading TLM should be were extremely high, and they were not being met.

But when I started to review the trading results via Tradervue and started looking at the results from various perspectives (not simply total % gains), there was definitely a high level of consistency that needed to be respected, even when the autotrading execution was less than ideal.


The final sense of a-ha was when I ran the trading results through a significance test (one-tailed t-test where mean P&L > 0).  I learned the method of determining whether you have an edge from Adam Grimes' book "The Art and Science of Technical Analysis" within chapter 12 (highly recommended, it's a keeper).

Even with all the slippage and missed trades, the p-value was 0.07 for The Lincoln Million autotrades, which means there's a 7% chance the results are random.  Lower p-value indicate the results are more statistically significant (better) and p=0.05 is a commonly used guideline for a significance level. 

This is an indication that even with all the challenges from autotrading, this system still has a solid edge based on the current data set.  

Assuming future improvement of missed trades and/or reduced slippage, or perhaps greater success from manually executing the trades, the statistical significance can potentially drop well below the p=0.05 significance level, further reinforcing the notion that this method has an edge.

And what if I take his published results over the same period and perform the t-test?  The resulting p-value is less than 0.01, which means there's less than a 1% chance his results are random or due to luck based on the particular set of data analyzed.  Doug's trading method definitely has an edge.


Perhaps the The Lincoln Million autotrading doesn't align perfectly with my personality (but then again, would any system you did not create yourself?), which is why having it autotraded without my biases or intervention has worked relatively well to date.

That's one reason why I'm approaching autotrading as a way to diversify my overall trading portfolio by 1) utilizing historically profitable trading services that 2) I couldn't execute consistently on a long term basis.  In essence, it's like I'm selecting and hiring traders with various trading methodologies and styles, and having them each trade a portion of my account.  And if I don't like how they're performing or how their performance blends with the other traders (services), I can easily "fire" them.

 So far, here's one of the biggest benefits from autotrading The Lincoln Million -- I have continued to track and follow the trade alerts with a much greater level of focus and involvement since I have "skin in the game."  I believe that has enabled me to learn this method much faster vs. a passive approach, and to discover certain Lincoln Million setups that resonate very well with my personality.

  • Since I only have around 30 trades for observation, which is barely the minimum data set required for a statistically significant test, I will continue tracking the system to see how well it continues to perform
  • Continue to monitor the number of missed trades/slippage to determine how much of an impact it has on overall results
  • At a certain point, I might consider manually executing the trading signals instead of autotrading to see whether I can obtain better execution, but having the time to trade is currently a constraint
  • Keep track of what happens if a trade is closed out when it's at a loss > $400 (based on current trade allocation level) vs. letting it ride hoping for a recovery.  This will be easy to track within Tradervue using the tag feature and the MAE metric
  • And finally, this exercise has reminded me of how I need to step up my trade journaling.  I know the more I leverage the power of Tradervue, the faster my trading will improve


Greg Reinacker said...

FYI, Tradervue now shows the p-value in the statistics tables - it's called "Probability of random chance". At the moment it's only visible for subscribers on the gold plan. Thanks for the idea!

Grove Under said...

Hi Greg,
Thanks much for the update and quick turnaround!


Anonymous said...

Any updates on the service?
Im thinking of joining


Grove Under said...

Hi Joe,

Sorry about the delay, but in case you're still looking, yes, Tradervue is still highly recommended. It keeps getting better and better all the time.

Most importantly, it helps to provide great insights to your trading via analysis of your actual (or SIM) trades.