Tweet
Hey,
I'm working with a client to analyze historical data for NBA matches and I managed to do all the data processing to get Moneyline odds over time for each game. The issue that we've identified is that odds in the early stages of the games are extremely stable for almost every match, even when the actual match had surprising upset scores in the first half, for example, which should have moved prices a lot.
I've attached a few example of odds (last traded price) vs. minutes after a match went live for NBA matches that show this behavior. The extreme stability of the prices in the first portion of the matches doesn't seem realistic to us, and my client has actually captured live odds for similar matches today that show much more volatile behavior, comparatively.
I don't think there's anything wrong with the way I'm processing the odds as it's pretty straightforward. I identified the ID for moneyline odds then track the information on the last traded price for each team over time as calculated by the timestamp on each record.
I'm working with a client to analyze historical data for NBA matches and I managed to do all the data processing to get Moneyline odds over time for each game. The issue that we've identified is that odds in the early stages of the games are extremely stable for almost every match, even when the actual match had surprising upset scores in the first half, for example, which should have moved prices a lot.
I've attached a few example of odds (last traded price) vs. minutes after a match went live for NBA matches that show this behavior. The extreme stability of the prices in the first portion of the matches doesn't seem realistic to us, and my client has actually captured live odds for similar matches today that show much more volatile behavior, comparatively.
I don't think there's anything wrong with the way I'm processing the odds as it's pretty straightforward. I identified the ID for moneyline odds then track the information on the last traded price for each team over time as calculated by the timestamp on each record.
Comment