About the data being suspicious (if that's the case), I'd really appreciate if someone could share better data. About opening gaps, I don't see how could that be relevant when I'm only trading based on close prices. As long as close prices are correct, this backtest should be correct.
I simulated this from 1962 and it looks very similar to what you showed. This is using TC2000 EOD data.
Some of the lower performance might be because of the faster, cheaper trading available now. Many times the market moves up a large percentage quickly after a drop. So buying at the close has lost most of the move's gains.
“If today's close is higher or equal than yesterday's close, buy/hold SP500.” 1. You cannot buy based on today’s close price after you know today’s close price. Close price isn’t known even when the market closes, and may vary substantially from expected closing price after the closing price is published a few seconds to minutes after market closes. Orders for the closing auction are submitted all day, while the closing price is reported after the market closes. The closing price may vary substantially from what it was if you’d issued an order before the market closed. Also, there is a cutoff time for submitting closing orders. 2. Exchanges underwent many regulatory and technological revisions through time. 3. The volume has substantially shifted towards the closing auction through the years. From a few % to now almost 30% at close on NYSE, for example. 4. Almost all basic and published strategies stop working after being found or published, for obvious reasons. And especially in the age of electronic markets and computers where everyone can backtest millions of strategies and start utilizing them on large scale immediately. Now almost every month we read about hedge funds whose strategies have stopped working after many years of being successful, and put them (and each other) out of business. So such basic strategy has absolutely no right to be still working.
Makes sense to me that electronic trading would smooth out this edge. Yeah it's humbling as a trader to realize that your edge could evaporate at any moment. It could also come back. CTAs run trend following strategies on commodities that go years and years without profit when the commodities are not trending.
SPX and the SPoos are two completely different trading "animals" and btw the S&P futures (large contract) didn't exist until 1982. And the eMini version not till 1997. And as far as I have seen, over time through bull and bear markets, down days are better for then going long.
See: How Markets Really Work/A Quantitative Guide to Stock Market Behavior by Larry Connors & Conor Sen. My results (and many others far, far sharper than myself) before reading the above, and since, are along the same lines.
Markets never ... should, will, must ... do anything. It is always about probabilities. And looking at a one day move in isolation is not high probability. (Although early on today it did go up) The more days in a row in one direction (also higher highs/lower lows) increases odds for a reversal.