Storage Markets Ends Experiment with Prediction Markets
Storage Markets has concluded our evaluation of using a prediction market technique to predict trends and transitions in the storage industry. The results of this experiment suggest that active prediction markets do appear to provide good quality forecasting results that are quantitative in nature and free from personal bias. If you are interested in learning more about our experience with prediction markets or our forecasting results, feel free to contact us via info at storagemarkets dot com.
Final Metrics:
- 550 days of trading
- 560 traders representing all corners of the storage industry
- 46 market research questions asked
- 200 securities traded
- over 1.5 million shares traded
- r-squared of 0.9422 (1 being perfect)You can find the last Market Wrap, our occasional summary of interpretations of the current trading results here.
Here’s the final, all time Top Trader leader board. Congratulations to wibni, Euchinie and rasrazz
We truly appreciate all of the storage industry professionals and end users who contributed to Storage Markets. You formed a unique online community whose collective insights created valuable forecasts to help make strategic decisions. We hope you found the trading experience both interesting and enlightening.
Thank you,
John [John Ives] & Rich [Rich Pappas]
November 09, 2007
No business model + bad choice of software provider (HSX).
—
UPDATE: John Ives comments…
HSX software certainly has a lot of head room to offer to a relatively small project such as Storage Markets. Given the prediction market software vendors available to us during late 2005, we were very happy with our choice of HSX. Alex and his team at HSX provided excellent customer service and were extremely supportive of our experiment.
Storage Markets, both the site and the business, were created to test a couple of specific business models based on prediction markets. Neither model appeared to yield the required ROI. Hence, we ended the experiment and returned excess capital to our investors.
—
External Link: John Ives’s blog
HSX software certainly has a lot of head room to offer to a relatively small project such as Storage Markets. Given the prediction market software vendors available to us during late 2005, we were very happy with our choice of HSX. Alex and his team at HSX provided excellent customer service and were extremely supportive of our experiment.
Storage Markets, both the site and the business, were created to test a couple of specific business models based on prediction markets. Neither model appeared to yield the required ROI. Hence, we ended the experiment and returned excess capital to our investors.
Holler ballers, long time no argument.
Just to repeat a point I have made in the past, prediction markets are not effective because they aggregate information from a certain group of people in a somehow more efficient manner.
They are more efficient when they are allowed to cast a much wider net, to reward predictions and people outside the known inner circle of “professionals” who can bring new information to the “market” (discussion).
PM models that insist on pre-defining the population of traders (because of fear of lawsuits, general risk aversion, etc) are doomed, because they don’t add any efficiency to the process. You have to allow anybody to join, or else you’re simply redefining an old discussion in “prediction market” terms without making it any more efficient. In fact, chances are you’re making it significantly less efficient.
This was why I was so incensed with the nobel laureates’ letter to the CFTC ages ago, because it cornered pools of liquidity in “non-profit” institutions under the asinine assumption that they would somehow get off the ground individually. No matter what the market, liquidity is an economy of scale. If you carve up liquidity among lots of small populations you add zero value.
The academic context of prediction market debates displays a usually worrying, occasionally stunning lack of understanding of liquidity. As long as the academic frame insists upon what amounts to grafting physics equations onto market processes, under the ridiculous assumption that “all markets are efficient” (regardless of liquidity conditions), prediction markets experiments will continue to fail.
I have nothing to say about how good or bad HSX was, but the general model on which the PM industry is based seems very flawed to me. If you can’t have public prediction markets, you might as well not bother. If you can’t have a consistent basis of rewarding smart predictions (ie a REAL MONEY market), then theoretically, you will just get a herd vote. And yes, I’ve read the literature on play money versus real money. I have yet to see how you get “long tail” participation without allowing for commensurate rewards.
I haven’t been this blunt before, because I have always been treated respectfully by the other members of this community, be they on Wall Street or in august academia, and I know a lot of members are a lot more invested in how the PM industry is perceived than I am. But I think this is a very important point that is not made, because people wedded to the industry on a career basis do not want to undermine the near-term perception of their industry, and academics do not want to admit that they were wrong, or free up the economic rent to private institutions which could be much more effective than academic-controlled prediction markets ever could. As long as the current status quo stands, disappointment is inevitable.
I am not in a position to address the multiple points Alex makes in the previous post. I can provide additional data points based on our experience for the reader to consider relative to Alex’s points.
Storage Markets was, what we called, a “semi-private” market. We limited trader registration to those who could convince us, one way or another of their involvement with the data storage segment of the IT industry. Our goal was to recruit traders from the entire value chain in our industry. This value chain spans semiconductors, systems, enterprise software, system integrators, distributors, VARs, analysts and importantly, end users.
The result was a very heterogeneous, diverse community of traders from all corners of our industry. The diversity of the community was crossed with relatively diverse market content. We believe this resulted in a broad set of traders and their respective insight on any particular question.
We regularly revisited and reevaluated our decision between semi-private and public market operation. The marginal improvement in the perceived benefit of a public market operation was contrasted against the cost. We could never convince ourselves and others that the expected incremental improvement in market quality was worth the cost at that time.
The perceived cost had two major components. First, the given the business models we were exploring, our corporate and media customers and partners were (right or wrong) very sensitive to the demographics of our trader community. A true public market would have reduced our customers’ perceived quality of our value proposition. I suspect, over time, this objection could be overcome as our customers became more comfortable with prediction markets.
Second, the discussion forums and chat rooms associated with our semi-private market had the characteristic of having a very high signal to noise ratio. The content generated by our community added a significant about of information to the market. We observed that public markets appeared to have a much lower signal to noise ratio in their various discussion forums. We were not interested in reducing the value of our discussion forums.
Lastly, I cannot comment on the general case and Alex’s post. I offer the above as a real world data point based on our experiences.