Time Spread Backtesting 2022 Q1 (Part 5)

I’m in the process of manually backtesting time spreads through 2022 Q1 by entering a new trade every Monday (or Tuesday).

I left off describing the controversy of managing time spreads. Simply with regard to adjustment timing, I have mentioned:

• ROI% thresholds
• Repeating ROI% thresholds if PnL “recovers” (and the need to operationally define that word)
• TD
• Drawdown %
• Using the spread strike prices as adjustment thresholds

With SPX at 4474, trade #7 begins on 2/7/22 at the 4475 strike for $8,041: TD 46, IV 20.5%, horizontal skew -0.1%, NPV 367, and theta 47. This trade reaches PT on 24 DIT with one adjustment for a gain of 10.4%. Max DD is -12%.

With SPX at 4403, trade #8 begins on 2/14/22 at the 4425 strike for $5,528: TD 25, IV 23.5%, horizontal skew 0.1%, NPV 246, and theta 27. This trade reaches PT on 29 DIT with one adjustment for a gain of 12.4%. Max DD is -10.3%.

With SPX at 4292, trade #9 begins on 2/22/22 at the 4300 strike for $5,698: TD 37, IV 26.5%, horizontal skew 0.2%, NPV 246, and theta 32. This trade reaches PT on 15 DIT with no adjustments for a gain of 15.0%. Max DD is -3.7%.

With SPX at 4366, trade #10 begins on 2/28/22 at the 4375 strike for $5,998: TD 37, IV 26.1%, horizontal skew 0.2%, NPV 261, and theta 35. This trade reaches PT on 15 DIT with no adjustments for a gain of 10.7%. Max DD is -3.5%.

With SPX at 4201, trade #11 begins on 3/7/22 at the 4225 strike for $6,598: TD 37, IV 34.1%, horizontal skew 0.5%, NPV 262, and theta 48.

With SPX up 1.04 SD over 11 days, the first adjustment point is hit with trade down 9%:

Trade down 13% after 22 DIT, but despite another scary look (see Part 4, first graph) this is not the second adjustment point:

PT is hit two weeks later with a gain of 11.3%. Max DD is the 13% shown above. Unlike trade #6, doing a pre-emptive adjustment this time is followed by max loss two weeks later (-20.1%).

I don’t necessarily think the base strategy is good or better just because trade management matters for trade #11. More than anything else, I feel lucky and I realize that at market pivots, I will feel equally unlucky going through other trades that proceed to lose.

While I await a larger sample size to make any determination about optimal strategy guidelines, I suspect this sort of ambiguity is what is what makes time spreads difficult to trade.

I will continue next time.

Time Spread Backtesting 2022 Q1 (Part 4)

Today I continue manual backtesting of 2022 Q1 time spreads by entering a new trade every Mon (or Tues in case of holiday).

With SPX at 4403, trade #5 begins on 1/24/22 at the 4425 strike for $7,598: TD 29, IV 25.7%, horizontal skew 0.2%, NPV 334, and theta 43.

Profit target is hit 32 days later with trade up 15.82% and TD 20. Nothing to see here.

With SPX at 4515, trade #6 begins on 1/31/22 at the 4525 strike for $7,578: TD 32, IV 20.8%, horizontal skew -0.1%, NPV 356, and theta 41.

According to the base strategy, this is not an adjustment point with SPX down 0.79 SD over 18 days:

Trade is only down 3% and TD (not mentioned in base strategy guidelines) = 4. Graphically, expiration breakeven looks close such that another substantial SPX move lower could mean big trouble.

This is exactly what happens the very next trading day with a 1.03 SD selloff. The pre-emptive adjustment just mentioned would have the trade down less than 9% and just under 14% the day after: an even worse selloff at 1.42 SD. The trade would then go on to reach PT with a gain of 14.1% after 30 DIT.

Sticking with the base strategy would put me at an adjustment point the next trading day (22 DIT):

Trade is now down 12% after a 1.03 SD selloff in SPX, which is not as bad as I might have expected. What is somewhat frustrating about adjusting for the first time down 12% is the next adjustment point lingering a mere 2% away.

The lesson here may be to mind the numbers (TD 4 is much better than 1-2) rather than a graph that may not be properly scaled for best interpretation. A holiday separates the two graphs above, too; the extra day of time decay can help PnL.

The very next day (23 DIT), that 1.42 SD selloff puts the trade down 16%:

The base strategy does not address days between adjustments, which means I should adjust on back-to-back days as needed. My gut wants to minimize adjustments because of slippage loss,* and skipping this second adjustment gets me to PT with a gain of 13% after 32 DIT. The risk of skipping the second adjustment is that a big move catapults me well beyond -16% to max loss before a second adjustment can be made.

Adhering to trade guidelines and making adjustments on back-to-back days results in a 10% profit after 37 DIT. For this particular trade, the specifics of management really do not matter.

You can get a sense of why I find time spreads so challenging, though. “The devil’s in the details,” and there are lots of details in which to get lost.

I will continue next time.

* — I need to trade more time spreads to get a feel for how significant this is.

Time Spread Backtesting 2022 Q1 (Part 3)

Today I continue backtesting time spreads on 2022 Q1 in ONE with the base strategy methodology described here.

With SPX at 4281, trade #3 begins on 3/9/22 at the 4300 strike for $6,718: TD 35, IV 30.2%, horizontal skew +0.3%, NPV 270, and theta 44.

The first adjustment point is hit at 13 DIT with trade down 7%:

One week later with trade down 9%, is this the second adjustment point?

The first adjustment was down 7% and trade recovered to -2% but never turned profitable. I didn’t specifically define “recover” in the base strategy guidelines, but in case it means “to profitability” then I should hold off until -14% for a second adjustment even though the graph suggests we may be on the slippery slope already. Without data from a large sample size I don’t necessarily think either way is wrong: I just need to be consistent.

Profit target is hit 15 days later with trade up 10.34%. Position is almost completely delta neutral at that point (TD = 358).

Thus far, I have covered three time spreads in 2022 beginning on Jan 4, Jan 18, and Mar 9. Let’s look at the rest assuming I enter a trade every Monday (or Tuesday in case of holiday).

With SPX at 4661, trade #4 begins on 1/10/22 at the 4675 strike for $6,528: TD 17, IV 14.7%, horizontal skew -0.8%, NPV 294, and theta 24.

On 8 DIT, first adjustment point is hit after a 2.44 SD SPX move lower with trade down 11%:

Second adjustment point is hit the very next day with SPX down 36 points and trade down 16%:

This adjustment is tricky with SPX at 4541 and current strikes at 4675/4575. If I roll to the nearest ITM strike at 4550, then I have spreads only 25 points apart. I could roll to an OTM strike that is at least 75 points away if not more. I could also close the 4675 spread and stick with one spread for now, but despite increasing TD, this also reduces theta ~50%, which could require staying in the trade much longer. One benefit of rolling to cut NPD by a target percentage (see fourth paragraph of Part 2) would be to eliminate this judgment call entirely.

Speaking of spreads only 25 points apart, I often see a recommendation to adjust (roll) time spreads when the underlying price moves beyond a strike. If the strike prices are extremely close, then another adjustment is almost a certainty and I would have to ask why bother even rolling into such a structure at all?

For our purposes here, I will roll to 4500 to leave at least 75 points between time spreads. That feels comfortable to me, but I have no data to support or oppose it.

Max loss is hit two days later on a 1.58 SD SPX move lower with trade down 20.1%. Staying with a market move of -2.2 SD over 11 days is going to be difficult no matter what the guidelines.

I will continue next time.

Time Spread Backtesting 2022 Q1 (Part 2)

I just got done watching King Richard. Like Venus Williams in the movie, we can’t win them all. We don’t have to win them all, though, to have tremendous success. I think we’ll find this to be true with regard to Q1 2022 time spreads.

With SPX at 4569, trade #2 begins on 1/18/22 at the 4575 strike for $7,648: TD 27, IV 20.1%, horizontal skew -0.4%, NPV 336, and theta 30.

Three trading days later, a 1.58 SD move lower gets us to the first adjustment point down 13%:

Adjustment may be done a couple different ways. Creating the new strike ATM will maximize theta but cut NPD less. Alternatively, I can aim to cut NPD by a target amount (e.g. 50-75%). Placing the new strike farther OTM in this way will boost TD at the cost of a lower position theta, which means more time in the trade in order to hit PT. Whether either is better depends on statistical analysis of a large sample size. That’s precisely why I seek the Python backtester.

This adjustment lasts just over one month until the second adjustment point is reached down 14%:

Moving ahead two weeks, is this the third adjustment point?

Unlike the second adjustment, trade is now up 4.6% with TD = 8.

I generally feel it more important to adjust when down money especially if moving onto the “slippery slope” portion of the T+0 curve. Such is not the case here. Another approach would be to trigger off trade drawdown rather than net PnL%. That is, if trade is up 8% and then falls to +1%, consider adjustment since it now registers a 7% drawdown from +8%. Again, whether any of these are better depends on statistical analysis of a large sample size, which is why I seek the Python backtester.

This particular trade is a winner either way. Without adjusting, PT is hit the very next day (49 DIT). With adjustment, PT is hit six days later. Adjustment costs me five days, but even at 55 days, a 10% profit is over 60% p.a.

I will continue next time.

Time Spread Backtesting 2022 Q1 (Part 1)

Ironically, while developing a Python backtester over the last few months (e.g. here, here, and here), I have completely gotten away from time spread backtesting. Today, I will revisit the manual backtesting realm by looking at time spreads in the first three months of 2022.

As seen in previous posts on the subject (e.g. here, here, and here), time spreads may be approached in a variety of ways. In the current mini-series, I will address a number of different details and tweaks. Rather than get confused, distracted, and drawn off course by manually backtesting one at a time, my ultimate hope for the Python backtester is to be able to algorithmically run through a large sample size of each variant and compare pros versus cons.

For now, my base strategy is as follows:

• Trade the nearest 25-point put strike ITM with 10% profit target and 20% max loss.
• Place short leg at least 60 DTE with the long leg one month farther out.
• Adjust if down over 7% on max margin by rolling half to the nearest 25-point ITM put strike.
• If trade is down over 14% on max margin, roll most distant spread to nearest 25-point ITM put strike.
• If PnL recovers and heads lower again, watch for -7% and -14% as subsequent adjustment points.
• To account for slippage and commissions, a $21/contract fee will be assessed.
• Exit no later than 21 DTE.

With SPX at 4799, the first trade begins on 1/4/22 at the 4800 strike for $6,688: TD 20, IV 10.6%, horizontal skew -1.1%, NPV 291, and theta 15.6.

The very next trading day, a 2.57 SD move down brings us to the first adjustment point with PnL -7%:

A 2.10 SD move down eight days later brings us to the second adjustment point with PnL -15%:

Max loss is hit five days later on a 2.44 SD move lower:

SPX cratering 2.31 SD in 14 days has resulted in a loss of 20.4%. Tough to overcome that! Although horizontal skew increases, it still remains negative while IV spikes ~90%. This suggests IV increase as a partial hedge in this trade.

I will continue next time.