When testing a model for trading I was advised by a trader to use “trade data” and not “quote data”. What is the difference?
“Quote Data” usually means top-of-book bid/ask data. (Also called BBO=best bid/offer, TOB, inside market, etc.) But more generally it can also mean all the current resting orders sitting on the book at any given time. These are limit orders that are less competitive than ones at the top-of-book.
“Trade Data” means the record of trades that were actually transacted.
Depending on your market/asset class, it can be advantageous to use one or the other or both. For example, in FX, the BBO data is usually what you get… in fact it’s very rare to find trade data in FX. And most charting packages will show only the bid, or only the offer. This makes sense if the trading venue is a dealing-desk type broker who charges a fixed spread: you always can compute the offer if you know the bid, and vice-versa.
But if your spread can change a lot, you may want to look at both the bid and the offer. This occurs in FX if you have an ECN type broker, and it definitely happens in futures markets. Usually the bid/ask spreads are widest just before a major announcement, or during off-hours. At such time you might want to avoid placing a market order.
In futures markets, I find it useful to compute most indicators from trades, and to use bid/offer data to filter out entry signals that will cost me a lot of bid/ask spread. And there are some indicators that take both into account… for example looking at how often the trades are occurring at/near the bid rather than at/near the ask is one way to try to predict a near-term move down or up.
Also, I like to do backtests on bid/offer data to get a realistic picture of my trade costs. It depends a lot on your trading platform, and what kind of fill models it offers. When I write my own backtests (in MATLAB for example) I’m able to create my own custom fill model that looks at the bid/ask price and quantity to estimate my “true” trade cost.
But one caveat here is to make sure your historical trades and quotes are synchronized. I’ve seen data (even sourced directly from the exchange!) in which the timestamps for the quotes were on a different clock than the timestamps for the trades. I was shocked. And I had to develop a way to sync up the historical data so I could do proper backtests.
As always, you have to consider the type of strategy you are trying to develop, and the factors that will make it win or lose, then decide what kind of data you need to use. And of course you have to look at your own skillset and budget to decide what is doable. You may just need to take what the platform gives you, and develop a strategy that will fit within its constraints. Or you may have the budget/skills/time to start top-down and build exactly what you think you need.
A monetary citation alludes to particular market information identifying with a security or item. While the term quote particularly alludes to the offer cost or solicit cost from an instrument, it might be all the more blandly used to identify with the last value which the security exchanged at (“last deal”). This may allude to both trade exchanged and over-the-counter monetary instruments.