Multichoice "Publication" Markets

On our COVID-19 preprint market, each preprint has 4 associated questions: 1 on Citations, and 3 linked Publication questions. This page explains how the linked Publication questions work:

For every preprint, exactly one of the following will occur:

  • No publication (within a year)
  • Publication in Low impact factor journal
  • Publication in High impact factor journal

Our platform allows you to bet separately on each outcome, each of which begins at 33.33%.


Splitting these up keeps the binary feel of our previous markets, and makes it easy if you mostly have strong feelings about one outcome. (Conversely, it takes more work if you have strong beliefs about all three.) For example, if you think “No publication” should be higher, and have no strong feelings about the other two, just increase the chance of “No publication”.

Because the outcomes are mutually exclusive, if you change the forecast for one, the others will move in the opposite direction so the three sum to 100 %.

I’ll demonstrate. In the multichoice market below, I am about to make the first trade. The market forecast begins at 33.33% on each of the three possible outcomes, but I am arbitrarily about to change the market forecast to 30% on “published in a journal with impact factor of at least 10” (a High impact factor journal).


As you can see, the “Related questions” show the linked multichoice questions (Low and No), and the independent Citations question currently at 50%.

After my forecast of 30% is submitted, the “Related questions” (below) is updated so that the other two possible publication outcomes have increased to 35%. 30% + 35% + 35% = 100%


With all our prediction markets, if you want to win, or help our research, your goal should be the same: find the market forecast that is “most wrong” and nudge it until you think a different question/market forecast is the most wrong.

Note, my bet/forecast above, of 33% -> 30%, was just chosen at random. But it might be a good bet if I thought that the true probability of this outcome was 20%. That would mean I only nudged the market forecast a bit in the right direction.

If I reviewed a bunch of other market claims, and couldn’t find another that I thought was “more wrong” then I could return to this one, and continue nudging it closer to 20%.

Or if I thought the most-wrong claim was now one of the linked 35% claims, I could click on those and adjust. (To get a feel for that, see “Getting a MultiChoice Forecast Just Right”.)

Happy Forecasting!

Mike Bishop

There is NO COST to participate in Replication Markets.

This research is supported (in part) by the Fetzer Franklin Fund of the John E. Fetzer Memorial Trust. It uses a platform developed for DARPA SCORE, and some staff are supported by SCORE while working on this. We are grateful for their support.