The fact that the gaps up each day went unfilled could suggest that a good portion of the buying may have been done by shorts who were forced to cover as prices began to get away from them. The number of instances has been quite low. On average we’ve only seen this setup occur about once every 2 years. In every case there has been a move lower in the next day or two. Statistics strongly favor the downside.
Note: for this study (and most others related to SPY gaps) I consider the gap to be from the previous day's close - not the previous day's high.