The following is research I posted for subscribers to PDS.
——————————————————————————————
After the ALKS debacle, we were left wondering, “Should we be excluding Pharma/Biotech stocks?”
After running several different tests, the answer is no.
We should continue to include Pharma/Biotech stocks. Since the difference in performance is slight, anyone who chooses not to trade the Pharma/Biotech signals is not likely to experience a significant under-performance.
It turns out that ALKS is the single largest losing trade of the Power Dip System over the last decade. We should be careful about over-weighting very recent history and under-weighting a much longer past.
The Baseline:
1% Risk, 10% Stop Model
For all tests, commissions of .01/share were included and the system was allowed to take as many trades in one day as cash allowed.
Now let’s exclude all the stocks in the Pharmaceutical/Biotechnology sector and run the same test:
Excluding the Pharma/Biotech sector eliminated 97 trades, and shaved about 3% from the annualized return.
What jumped out at me was the big differences in the Avg. Profit and Avg. Loss. This difference shows how volatile the Pharma/Bio stocks are. Those 97 trades were a small percentage (just 2.3%) of the total number of trades, yet they increased the Avg. Profit by 24.3% and increased the Avg. Loss by 25.4%.
Exposure dropped by about 1.4%, meaning that these Pharma/Biotech trades constitute a very small percentage of exposure to the system.
Let’s look at performance using ONLY the Pharma/Biotech sector stocks.
Surprisingly, using only the Pharma/Biotech stocks, the system maintains most of its original characteristics. The Max. trade % drawdown of -32.96% is from the ALKS trade. ALKS is the single largest losing trade experienced by the system during the test period.
Trading the Pharma/Biotech stocks, exposure was only 5.15%. Remember that normally these stocks are mixed in with others, so the system’s exposure to them is much less than 5.15%.
Now lets look at the worst and best 20 trades of the Pharma/Biotech stocks:
Note that the stop has been very effective. Only twice have Pharma/Biotech stocks gapped down past the stop. Also note that these stocks have offered significant gains.
In real-time trading there may not be any liquidity at the stop-market price, and therefore there is no guarantee that the actual stop-market price could have been obtained.
De-Listed Stocks in the Database:
While these tests include de-listed data, once a stock becomes de-listed, it loses all of its sector classification within the database. This means that there are Pharma/Biotech stocks included in the testing that are not being filtered out because they are de-listed.
To account for this, I ran a test using ONLY the de-listed data. I’m including the stats because there are not many systems out there that make money using de-listed stocks and I think the statistics are interesting.
The statistic do show some degradation of performance, but the system maintains profitability.
Now, lets see the worst and best 20 stocks from the de-listed data.
I have made red the symbols from Pharma/Biotech stocks.
On the losing side, we see many examples of stocks that gapped past the 10% stop. Be sure to examine the MAE (maximum adverse excursion). The MAE shows the maximum amount that the position moved against the trade. If the MAE is larger than the loss, it means that the stock gapped, the position was stopped out, and the stock kept moving downward. (AmiBroker records everything that happened on the day of the stop-out.) In most instances, the stop worked well to protect against steeper losses on the day of the stop-out.
Note the largest winner was a biotech and there were two other biotechs that gained near 30%.
Finally, most of the worst 20 were not Pharma/Biotech stocks.
Summary:
The Pharma/Biotech stocks can be extremely volatile. However, the system’s exposure to them is very small. The system metrics maintain their general characteristics, even when trading only Pharma/Biotech stocks. Because the system uses stops, most of the damage from a volatile losing Pharma/Biotech trade is mitigated, while the upside from a volatile winning Pharma/Biotech trade is unlimited.
Therefore, the system will continue to trade the Pharma/Biotech stocks.
Thanks for doing the legwork on this. As a discretionary trader, it’s something I’ve always wondered about.
With the benefit a couple of beers, umm, of 20/20 hindsight (read: “with the benefit of you having done the real work!”), your conclusion’s also got a reasonable handwaving argument behind it, and one I wouldn’t have believed had I not seen the data: any system based on tracking price/volume signals will necessarily pick up on the movements of biotech stocks during the period leading up to the binary event.
There’s no way to predict the outcome of a binary event. But if you play a game often enough, and the game has a positive expectation, you still come out ahead. The whole point of systems trading is to take the (emotional, fallible) man out of the loop. A dozen “trade now, sell before the binary event” signals — even if the system isn’t going overboard by trying to use options/implied volatility to tell a human that a binary event is approaching — can make up for the occasional horse nuking on either the long or short side.
(Hmm. We have “horse nuke” in the iBC dictionary on the downside, but BTFO qualifies only as the upside equivalent of the “homo-hammer”. It’s kinda depressing that we lack an upside equivalent to the “horse nuke”. Now you’ve got me wondering what was the most egregious premium offered since iBC’s inception… and/or why we haven’t included it in the dictionary.)
saw the title, didn’t read a single thing afterward. will note that it took a long time for me to scroll down to the comments section. this suggests i am missing some quality analysis from the esteemed sir woodshedder. (no sarcasm).
i don’t care. straight up, i will never trade a biotech company.
love your work, sir. love that you addressed this “issue” when i had been thinking it for some time.
should i go back into this post to see how the powerdip fared if i excluded biotech companies? doesn’t matter, i might already know from personal experience.
Yeah, this research doesn’t apply to just any way to trade biotechs, only to the way the PDS trades them, i.e., average hold time of about 5 days, etc.
Anyway, you should get a cup of coffee and go back and read it!
what program are you using to test the dip system with? Amazing work btw
M, I use AmiBroker. Thanks!
Great post woody.
Thanks B! How are you?
Very well, thank you. It been a good year for MR system traders I think, particularly the last few months.
Useful info – thanks, Wood!
Hi Woodshedder,
Could you please email me afl code for etf-rotational-system-v10-part-4 with a Moving Average Filter.
julia1240 at gmail dot com
Sorry for posting in the wrong thread. Discovered the power of Amibroker 2 days ago, used stockfetcher for couple of years.
Many thanks!
Julia, let me look for it. How did you find about it, if you don’t mind me asking?
Someone posted a link to fidelity rotational system on Dan Sheridan forum.
Nice analysis, Wood. I’m glad you laid to rest this question the biotechs’ fitness. I know it’s been bothering a lot of us. But, seeing as they’re good for the PDS (or at least not that bad), I feel a lot better about trading them.
Glad it helped Data! I feel better about it too. Knowledge is power.
Anyway, we are biased towards over-weighting the recent past. If I ran a similar report about stocks that trade below 5/share, it would probably be a lot more significant than this Pharma/Biotech test.
True; there’s probably a whole pile of sectors of stocks we could pick out and finger-point at (biotechs, lower cap stocks, the list goes on). Beside the fact that it would be a bit impractical to test for all those (just read your post where you said it takes you 5 hours to run one of these… eeks! didn’t realize that or I wouldn’t be asking you for tests all the time), but you also start to risk curve fitting. So it’s probably better, in this case, to “keep it simple,” and just take the good and the bad as part of life in the markets.
I worked in Biotech for many years. A Biotech has no control over it’s own destiny. It takes 15 – 20 years to go Phase IV, if you get there.
If the Phase I trials look good (mice don’t die) a few eyebrows are lifted.
Phase II FDA approval gets everyone moving. The scientist that came in a did some contract work for a few weeks all of a sudden wants a cut, his lawyers are shaking with excitment, easy money. The lawyer that defended the company pro bono in a minor lawsuit and won wants to wet his beak, he’ll sue also. A similar drug in the market makes someone die, boom, you’re dead.
After filing the BLA for Phase III the FDA auditors sends back the finding “data inconclusive or doesn’t show promise, we need more data”, which means another two years of trials….
Give me a company that can control it’s own destiny, that’s a company worth investing in….
About a month ago, using my system – an RSI2 mean reversion system – I lost 49% in one day on ARNA, a biotech stock that had a bad day presenting their latest drug to the FDA.
I’ve taken the biotechs out of my watchlists.
Hi-
I trade a mean reversion system, and my objective is most gain WITH LEAST PAIN. This can be expressed as highest CAR/MDD, highest Profit Factor, highest Risk-Reward Ratio, etc.
Looking at your first two data tables, CAR/MDD went from 2.14 to 2.09 (worse), Profit Factor went from 1.39 to 1.41 (better), Risk-Reward Ratio went from 1.05 to 1.09 (better). You also state “…they increased the Avg. Profit by 24.3% and increased the Avg. Loss by 25.4%” which implies that the extra gain was NOT worth the extra pain..
When using my strategy along with my data, excluding most pharma stocks, slightly improved CAR/MDD, greatly increased Profit Factor, and made Avg Profit/Loss better. Based on that, I decided to exclude pharma stocks.