Yes, this is a heated subject but I hope that the example from recent SPY data will convince you that there is no easy answer.
The other day I wrote a blog with the intention of demonstrating that traders often attribute timing ability to models that in reality have none and a reason for this is the dividend effect. However, this turned into a heated discussion about adjusted vs. unadjusted data in the blog comment section although that was not my initial intention. I was even accused of being ignorant in emails and that I do not account for dividends that buy cars, for example. I will post an example from recent SPY data below that will demonstrate the impact of data adjustments. But first, let me switch to John Orford mode for clarity, because talk is cheap and all that counts is numbers.
The most recent dividend payment in SPY was a $1.03 on June 19.
On the chart below, the top series is adjusted SPY data, the bottom series ($SPY) is unadjusted and the vertical marker set at the dividend day:
Note that in both series, data after June 18, 2015 are actual, unadjusted, because there is no more dividend payment (a little divergence form John Orford mode here).
Next consider a long-only, 5-30 crossover system that buys and sells at the open of the day following the signal.
Buy if ma(5) > MA(30)
Sell if MA(5) < MA(30)
Backtest starts on June 18, 2015. First the results based on adjusted trade series:
There is a first trade entry on June 23 and the exit is on June 30. Trade loss is -2.30%.
Note that these are actual prices. The trade is taken and booked. There is nothing you can do to change the results in the future.
Now results using unadjusted data:
The first entry is again on June 23 but the exit is on June 26. The trade loss is -0.87%.
You lose additional 143 basis points using adjusted data and your boss is very upset.
Both are actual trades. You can only change the results with backtesting hindsight.
Don’t worry, next time you may be lucky and you may make more with adjusted data. So ask your boss to forgive you.
That was the point though. For real traders there are differences.
What to use? I do not have an easy answer for real trading but in the case of testing models for significance, dividend-adjusted data can lead to wrong results.
Splits, and rollovers in the case of futures, have no effects if proper exits are used.
As I wrote in the comment section of the controversial blog the other day:
“If there are spits data must be adjusted but relative returns stay invariant under the transformation because C/Cn stays invariant as both numerator and denominator are divided by same constant. Therefore, backtests are good as long as they do not refer to point exits or absolute price levels.
On the other hand, in futures where the rollover introduces a fixed displacement of all prices in points, first differences stay invariant (C-Cn) and backetsts are good as long as they make no reference to percentage gains.
I have talked about the above in my out-of-print book in 2000 but the concepts are easy to understand if one is willing to do some test examples.”
Thanks to those who provided support during the controversy the other day.
You can subscribe here to notifications of new posts by email.
Charting program: Amibroker
Detailed technical and quantitative analysis of Dow-30 stocks and popular ETFs can be found in our Weekly Premium Report.
© 2015 Michael Harris. All Rights Reserved. We grant a revocable permission to create a hyperlink to this blog subject to certain terms and conditions. Any unauthorized copy, reproduction, distribution, publication, display, modification, or transmission of any part of this blog is strictly prohibited without prior written permission.