Look at the Order column in the feed. In the capture below, every visible row reads MARKET. That is not a criticism of anyone, it is what happens when a fast path exists and nobody has a reason in mind for using anything else. But a market order is a purchase, and what you are buying is immediacy. The sensible question before every entry is whether this particular signal needed any.
The answer differs enormously by which engine produced the row, and the feed puts the source right there on the table with a filter chip for each one. Once you sort your execution by source, a lot of small recurring cost disappears without changing a single thing about which trades you take.
Immediacy is a thing you buy
A market order crosses the spread and takes whatever depth is standing there. You pay the spread plus any slippage past the touch, and in exchange you get certainty that you are in the trade right now.
A limit order at or inside the touch pays less, sometimes nothing, sometimes it earns you the spread. In exchange you accept the possibility of not being filled at all.
So the trade off has exactly one variable in it. How much does this signal's edge decay while you wait. If the answer is "not much", the limit is free money. If the answer is "the whole thing happens in the first four minutes", then a limit that misses has cost you not a few basis points but the entire trade, and the market order was cheap at twice the price.
Reading half life off what each engine measures
Trade Alpha lists its sources on the same page as the feed, with a one line description of each, and those descriptions are enough to rank them.
One caveat before the ranking, and it is not a formality. I am reasoning from what each engine measures, as the platform itself describes it. There is no published half life statistic on the page and I am not going to manufacture one. If you disagree with where I have put a particular engine, the thing that settles it is your own fill data over a few dozen trades, not my argument.
The engines from fastest to slowest
MTE, the Momentum Trading Engine. Described as eleven entry systems across crypto and equities. Momentum entries are, by construction, triggered by something starting to move. The move you are trying to participate in is already underway when the row appears, and the cases where these signals pay are disproportionately the ones that keep going without pausing. This is the source that most often justifies crossing the spread, and the one where a resting limit twenty basis points below is a good way to watch the trade leave.
MRE, the Market Reversal Engine. Described as multi signal confluence for tops and bottoms. A reversal signal is a statement about a level, and levels are the one thing in markets that price tends to revisit. That is not a claim that the reversal will work. It is an observation that the entry zone typically remains available for a while, because a market turning at a level is by definition not sprinting away from it in the first minute. Reversal rows are the strongest case for patience in the whole feed, and paying for immediacy on them is close to pure waste.
AOE, the Asset Outperformer Engine. Described as cross timeframe relative strength. Relative strength is a ranking measured across timeframes. Something does not stop being a relative outperformer between eleven and eleven twenty. These rows tolerate a worked order comfortably.
MRS, macro tilts. The macro scorecard produces a posture, not a trigger. A tilt that is right is right for weeks. Executing it with a market order is paying a premium for speed on a decision whose horizon is measured in months, which is the clearest case of mismatched urgency the feed offers. Work these over hours, or over several sessions if the size warrants it.
MANUAL, desk entries. No inference is available. A manual row could be either extreme and the interface cannot tell you which. Unless you wrote it yourself and know why, default to the patient side.
What the spread is actually costing you
Use the prices from the capture to make this concrete rather than abstract.
Ford was quoted around 14 dollars on the visible rows. A one cent spread on a fourteen dollar stock is roughly seven basis points. On a 1,500 dollar position that is about a dollar. BK at 157.13 with a two cent spread is about 1.3 basis points, or twenty cents on the same position size. Neither number will change your life on a single trade.
The arithmetic that matters is the accumulated version. Four entries a week at seven basis points on 1,500 dollars is four dollars a week, a bit over two hundred dollars a year. If half of those entries were reversal or macro rows where waiting cost you nothing at all, then roughly a hundred dollars a year was spent on immediacy that no signal required.
Crypto is where this stops being small. Spreads on thinner pairs run wide enough that a market order on a 1,500 dollar clip can cost thirty or forty basis points, six dollars a trade, and the same discipline applied there is worth several times what it is worth in liquid equities. The silver row in the capture, and any commodity or metals row, sits somewhere in between and deserves a look at the book before you decide.
Where the rule breaks
Three exceptions, and they matter more than the rule does because each of them is a way to lose real money while feeling disciplined.
The first is thin books. On an instrument where the spread is wide because there is genuinely nobody there, a patient limit is not patient, it is absent. Worse, a resting order in a thin book is a piece of information you have handed to everyone watching. In that situation a small market order, or a limit crossing the spread deliberately, is the honest choice, and the real answer is often that the position should be smaller.
The second is exits on risk. Everything above concerns entries. A stop is not a place to be clever about the spread. If your stop level trades and you are working a limit for a better fill, you no longer have a stop, you have a hope. Risk exits go at market regardless of which engine produced the entry, and the few basis points you give up are the cheapest insurance in the account.
The third is gaps and news. When something has just happened, half life collapses for every source at once, including the slow ones, because the reason the row exists has changed. A macro tilt row that lands ten seconds after a central bank headline is not a slow signal any more. If you cannot explain why the row appeared when it did, patience is not a strategy, it is an assumption.