Selling short put options awesome backtest data

This vid describes outcomes of 41k trades selling put options backtest on SPY – 10 years of data. The targets were 45 days out at .16 deltas. They tested closing out at profit targets of 25, 50, 75% and expiry. Closing of losses was 100, 200, 300 and expiry. It’s very informative, data rich, and well-presented with charts. It’s difficult to give easy take-aways, but here’s a couple of screen shots to whet your appetite. Go watch vid, though – these guys did a great job.

Win Rates on Short puts with various parameters for closing

Win rates for short puts

High success rates for all trade parameters. Colors are loss targets (green is 100% close), with profit targets moving from left to right in each color (left-most is expiry, then 25, 50, and 75%).

Win Rate adjusted for breakeven

Basically, this is the win-rate excess of that required for profitability. Taking 100% losses and 25% profits gave the best outcome here.

Average PnL

The data is ordered a bit oddly, since the left bar in each colour is letting any Lossward moving trades (stock went down) expire. It would make more sense to put them at the end of the color set, but no matter. This shows that taking smaller profits has the lowest profitability, unsurprisingly.

Time held adjusted average PnL on selling put options

This shows the PnL over time used. Since the first green bar holds everything till expiry, it’s pretty low. Similarly with the first bar of every colour: since this holds profit till expiry, and the win-rate is very high at .16 deltas, most of these trades expired.

Max Drawdowns by strategy for short puts

These are drawdowns. Taking losses obviously reduces max drawdowns in a crash.

Selling put options PnL by Implied Volatility (VIX as proxy)

Here is results further broken down by vix levels. VIX is used as a proxy for volatility. It’s a little imprecise and can be a lagging indicator, however, it works well enough. I would have used ATM Implied Volatility because that’s really targeted to premium. At any rate, no surprise- higher volatility leads to higher PnL. Taking losses rather than letting losses expire was much better.

Drawdowns by volatility

I love this data. Selling put options backtesting allows huge amounts of information and it’s tough to sort through. These guys did a great job showing useful data here.

  1. Basically, in a low VIX, low IV environment, it appears you may not want to set loss targets. BUT – appearances are deceptive. That’s because the trade starts in a low-VIX environment, then the market crashes and moves into a high VIX environment. But you’re entering the trade low-VIX, so don’t get suckered there. That’s actually the worst because you get the least premium and get hit the hardest. So, it’s actually definitive that high-VIX would result in the largest drawdowns.
  2. The real interest is the ginormous difference in taking losses versus leaving them open.

Time adjusted PnL with volatility breakout

So this is pretty cool. High volatility with 300% loss target and 25 or 50% profit target gives excellent time adjusted PnL. However, with their parameters, once you exit the trade, you won’t enter the next until monthlies arrive at 45 days out. So, with a 25% profit target, you’ll probably only stay in the trade for 8 days or so on average. Your monthly profits will be substantially less, because you aren’t entering a new trade until week 2 (45 days before the next monthlies). I don’t see any reason to wait for this with the number of weekly options available.

It would make sense to wait until the market drops, though. Always better to enter a short put trade when the underlying is off-peak.

Conclusion

I recommend checking out the video if you want to see selling put options backtested. It’s good data, well-presented. It would take a fair amount of massaging to develop a trading strat out of it, but it would definitely help. I’m trading weeklies right now and entering the position early on Friday. This creates an accelerated time decay. It’s not huge because much of the weekend time decay is already happening on Thursday.

Anyway, look it over and I hope it helps. Thanks to the ProjectOption guys for such a great data set.