A Momentum Strategy In SPY During The Last Hour
Last Updated on November 21, 2020 by Oddmund Groette
Here is the hypothesis: if SPY rallies from the open and until the last hour, can we make money by going long one hour before the close and exit on the close?
Previously I have written about the different periods of the trading day:
- At what time of day does SPY set high and low?
- SPY sets low or high in the last hour – a daytrading strategy
- Return in the first hour, middle of the day and the last hour
This is a strategy I have traded live over the last year (just a few trades). I have tested with different thresholds, but it seems like the more SPY is up from the open (it works better from the open than from yesterday’s close), the more likely to rally the last hour. I have used a threshold of 1.25% from open to the last hour, and I get 18 fills from 2010 until the present and 13 winners. This is an average of .39% per fill, not bad for such a short holding period. Considering a total holding period of 18 hours over more than 2.5 years, a total gain of 7% must be considered extraordinary. Here is the equity curve:
There are few fills, though. Increasing the threshold to 1.5% we get 8 fills and 6 winners, and the average increases to 0.49%.
If I measure from yesterday’s close to the last hour, I don’t get such remarkable results. I’m not sure why. Does anyone have any suggestions? It’s still good, but average drops. Here are the results from yesterday’s close to last hour:
Threshold | P/L | #fills | #wins | Avg |
1.25 | 7.41 | 66 | 40 | 0.11 |
1.5 | 4.89 | 42 | 23 | 0.12 |
1.75 | 5.53 | 29 | 19 | 0.19 |
2 | 4.07 | 21 | 14 | 0.19 |
2.5 | 3.89 | 12 | 10 | 0.32 |
3 | 1 | 4 | 2 | 0.25 |
To reverse it, ie. to go short if a large drop from the open, does also work (see below for explanation). For small movements, there seems to be a tendency downwards, but large drops get reversed. However, measuring from close until the last hour my results are a bit disappointing. Using from the open I get tradeable results using a threshold of about 1.5%, but 1.75% seems better (the more movement the better). Here is the equity curve for 1.75% (15 fills, 8 winners):
Why does this work? I have no idea except for the following:
- Short covering, or
- rebalancing in ETFs
There are a lot of traders who fade SPY. If SPY just continues higher, a lot of them might want to take a stop-loss by covering and thus needs to buy (and they may take profits on big down days and buy during the last hour). The other reason is the growth in leveraged ETF’s. A lot of the leveraged ETF’s needs to rebalance every day and this pushes more fuel to the fire. You can find a description of this here. When the market goes up leveraged bull ETFs need to buy more to keep the leverage ratio. When the market drops, it’s the opposite effect, because bear ETF’s need to buy. However, this is not my field so I strongly urge you to read this research paper and this blog (thanks to Rob Hanna and ATrader for making me aware of these articles). According to the mentioned research paper, as much as 10-50% on the MOC orders could be attributed to these ETF’s! Reading the research paper gave me new insights into how the markets work.
Larry Williams once spoke about “The trade nobody wants”, which is buying at the low of the day on the close. From practical experience its a bit same if you are 1 hour from the close in an uptrend, as a short term trader you do not want to take that (long) position. Quite often though it leads to a close rally. Its quite logical – however with the change of markets I am uncertain how much importance should be put on MOC going forward.
Hi Oddmund,
I am not sure what you mean by “According to the mentioned research paper as much as 10-50 on the MOC orders could be attributed to these ETF’s! Reading the research paper gave me new insights on how the markets work.”….
It’s supposed to be 10-50%.
Hi,Oddmund:
Nice discovery. Just wondering where you get intraday data for backtest.
Thanks.
Hi, you have to buy it – pay for it.
Thanks. Which data source would you recommend?
I have used IQFeed.
Ho….this strategy still works? Do you know the drawdown?
I’m trading it, but few trades. I have not checked its performance since 2013….