This would be an "aggressive" strategy. You get a signal from some source (looks like the author worked in news events, but the signal could be anything). If the signal indicates, via your model, that the product you are trading will go up through the best offer, you send an aggressive buy into that offer.

If all works out well, you get filled for a buy and the old offer (which you bought at) is now at least the best bid. Do this hundreds of times a day and you're starting to talk about real money, especially since you can scale (to a point) by upping your quantity.

Sometimes you aren't fast enough though. That buy you sent into the best offer is too slow - it merely appears as a buy at what is now the best bid (or doesn't at all if you use an IOC - immediate or cancel), and you get no trade. Even if you only get filled on 20% of the signals, if they are on average good, then you still make money.

This is just one type of strategy though. Not all HFT strategies rely on aggressive signals, though a lot do.

Just making a guess here but it may be that you have several limit orders fail to be filled for any given trade but when you you place an order that finally does get filled it may still complete a winning trade despite being your 5th order placed for the trade.

buying the bid without it going offered there. Then you sell the offer and have earned the spread.