A Look at a Real-Time Fund Trading Systematic Models
Two of the common things that I hear when “non-believers” review the performance returns of systematic investment models is “that’s just hypothetical” or “that’s a backtest”. I get it. Sometimes the numbers look to good to be true.
In a moment, I will share with you the results of a live, real money account, but before I do, let’s review.
It is agreed that a model’s inception often begins with testing ideas and then ultimately a backtest. That is the nature of model building. As a general rule of thumb, all model ideas that I develop must be grounded in academic research and the efficacy of the approach must have been clearly demonstrated. Only then do I become interested.
“What Works on Wall Street” or more recently, “Quantitative Value” are excellent examples of books that can be mined for investment concepts that are both well thought out and well tested. There is also the content to be found in the numerous white-papers that can be found at sites like SSRN.
As a general rule of thumb, the simpler the model, the better I like it. Moving average systems such as outlined in Meb’s paper are about as simple as you can get but the approach works by managing downside risk and thus positively impacting long term compounding.
The use of momentum is hardly new. There are countless papers and references in the literature discussing its efficacy across all asset classes. There are hedge funds and large institutions managing billions of dollars around momentum strategies.
Take a simply moving average system and add a relative momentum ranking and you’ve got the basis for a solid model that can be applied across just about any portfolio. If you want to get extra fancy, you can add an absolute return criteria and perhaps some measure of volatility but my point is that a successful system does not require hard to understand or arcane rules. In fact, I think the more specific and complex a model becomes, the more likely I believe it will eventually fail in the future.
When it comes to stock portfolios, the above cited books display quite clearly that factors such as value, yield, and momentum have consistently been attributes of stocks that beat the market. Intersect all three attributes and you’ve got a winner.
Models work because they remove human emotions. The reason Dalbar has been reporting for decades that individuals underperform (by a lot!) the indices is because of no other reason than human behavioral flaws. It’s not our fault…it’s how we are wired…it’s nature. I am sorry to be pragmatic but it’s also why I believe that the Dalbar report will never show anything different.
To overcome this deficit…we use models because models are concise. Models are consistent. Model’s do not care about the “story”. Models only care about statistical and factual data that can be accurately measured.
So now back to topic.
During the past year I was asked if I would be interested in managing a fund that utilized the exact models you will find on this site. I agreed. An account was established at Interactive Brokers and was initially funded with multiple six figures. I mention that only so readers understand that there is real skin in the game and real accountability. I also mention that so that you can get a sense of my order size which ranges anywhere from a few hundred shares to a few thousand shares. In other words, not too different from what I would imagine is the order size of most readers.
So at the beginning of the investment process I created a blend of the following models.
- Smallcap Fiscal Momentum
- Smallcap Rockets.
- Dividend Value
I more or less evenly allocated the capital across these models and began to follow the exact recommendations generated each Monday morning. So in other words, I was doing exactly what I recommend that our members do. I was also eating my own cooking as all investment newsletter editors should.
In February I introduced a MidCap model to the fund that used many similar factors. This was done to improve trading liquidity.
Here now are the results I promised earlier:
YTD return: 9.98%
At its peak in early April, the fund was up about 12%.
In the same period a blend of small and mid caps (50% IWM and 50% MDY) is up 0.35%.
I am happy with the fund’s start for several reasons, aside from the obvious positive performance differential.
What I am most happy about is that I think this demonstrates that trading model signals can be translated into real-time trades. Aside from three stocks where I was uncomfortable with the bid/ask spread size or I felt the volume was insufficient, all trades were taken.
The dollar weighted returns from the “published” model returns as compared to the “actual” real returns are almost the same. This suggests that investors can indeed take a published model and reasonably expect to achieve similar results even with slippage and transaction fees considered.
Second, I think the good news for individual investors is that it demonstrates that there is still alpha out there if you are willing to exploit stocks that are either misunderstood or simply not followed as in the case of many small caps.
Proponents of “efficient market theory” will tell you that everything is already priced into the market…to invest passively. But our results and the results included in the books such as “What Works on Walls Street” will tell you differently…that market beating returns are possible when using a systematic and disciplined approach that compound factors leading to positive expectancy.
I realize that as the fund grows that I will be eventually forced to increase market cap to maintain liquidity and some of the “size” factor I am exploiting now will diminish. But even in that case, models such as Dividend Value where the average market cap is well over $1 Bln, continue to display a high degree of positive expectancy.
In the coming weeks I will be exploring signal to trade automation. I am also tilting my bias to large-caps given the recent underperformance of small caps.
I will keep readers updated on future performance, but for now, let’s agree on one thing – this is not just hypothetical.