I am also interested in this field, I saw a video of a prop trader saying 90% of their trades are selected by looking at bid/ask and order flow. The only thing is to get decent returns you need to leverage to the hilt. I have been trying it out all week, just looking at the bid/ask and gauging the path of least resistance and acting on it.
This is actually how I learned how to trade - by scalping Bond Futures. The biggest rule for that endeavor was that if the market was 2/32'nds against me - get out, even if it meant hitting a bid or lifting an offer. I would also watch the boards to spy on the cash OTR Treasuries markets trade and also keep an eye on relevant coralaries at the time - namely, the S&P and the Yen futures over at the CME. It's very important to understand why there might be buying or selling pressure. IMHO it's more art than science. You also have to develop a keen sense of anticipation, and you certainly can't let a loser run. I would imagine that for the major name futures it can be tough in the sense that there is quite a bit of order cancellation and shuffling in the order books, and it seems to me at least that quite often the entire best bid or offer trades out in a fraction of a second. This is developed tradecraft - paid for in blood, sweat, tears, and account equity. If it was as easy as ferreting out the pure science or analytics and getting a fast ECN, then it would be impossible for a manual point-and-click trader to do it consistently. As it is, it's just next to impossible.
How far does a loss have to go for you to consider it a bad trade ? How do you filter noise from an actual directional move against you ?
If you had read my post more carefully, I addressed both those questions at least as it related to my own personal experiences in the CBOT 30 yr Bond pit in the early 1990's.
Even the DOM traders I know of pay more attention to recent volume of actual trades on the tape, than to the book, because of fakes and what not. Personally I think it makes sense to look at where people actually just placed bets (which they'll eventually need to get out of), vs looking at what may or may not materialize depending on where you "feel" icebergs and fakes may lie. Then again, maybe hunting for those surprises is a fun experience in and of itself; I wouldn't know. The guy from Jigsaw Trading, what's his name... ah yes, Peter Davies. I like his keep-it-simple approach to intraday DOM and volume.
He's done a bunch of video demonstrations, most on live markets. One which literally is about keeping it simple is:
Charts are a visual illustration of what actually happened. Charts are like footprints. Once you learn to recognize, say, elephant footprints and you see a herd's worth of them pointing in the direction of a well-known watering hole, chances are pretty darn good that if you follow those footprints you'll reach that watering hole. Now, maybe some unknown obstacle will divert the herd and prevent it from getting to the watering hole and by following those footprints you'll end up lost, thirsty and hungry. But way more often than not, the elephants get there. Same with trading off charts. You see certain footprints pointing in the direction of a well-known resistance level. These particular footprints tell you the majority of traders believe price really wants to get to that level. Now it's possible anything can happen along the way that could interfere with the belief price will get to that level, but more often than not those footprints lead to price heading at least to that level. If the orders you see on the book were not allowed to be pulled, then they'd be useful. By the time they're executed and you see what was real, the opportunity to put on a trade with the R:R required to net profits over time might be long gone. That said, I find unusually large size at a price level on the bid or offer tends to act as a magnet.