A better source than bitcoinwisdom for next dif. estimation?

[fairlay]

Some time ago we did a comparison of different sites that estimate the next difficulty: https://bitcointalk.org/index.php?topic=592552.0

Back then https://bitcoinwisdom.com/bitcoin/difficulty was the best source.

Lately we have noticed that user on our website probably have access to a better source. We offer predictions on the next difficulty (basically you can bet on it against other user https://www.fairlay.com/event/category/bitcoin/difficulty/). However, in the last days bitcoinwisdom always estiamted the next diff way > 27B. But on our site a lot of user has been willing to bet on "no" even with odds that would indicate that "no" is more likely (thus not in their favor). Now it turns out that the estimated difficulty indeed sunk on bitcoinwisdom and it is now only slightly above 27B.

This leads to the conclusion that their might be a better source for difficulty estimations - any idea what it is?

[Grix]

Common sense.

The best estimation source is your own brain. Look at the difficulty fluctuation trends for the last few weeks and combine it with the raw values since the last adjustment and you should get a very good picture. The reason bitcoinwisdom was off this adjustment is because it uses the last difficulty adjustment as a variable in its calculations, and the last adjustment was clearly a statistical anomaly, way higher than usual. Also, the difficulty had a sharp spike right at the adjustment. Usually spikes like this is due to random chance and subsides a bit afterwards, as it did this time too. Knowing this, and combining it with a quick look at actual graphs should have made people know that bitcoinwisdoms prediction was useless.

[runeks]

You should create an instrument that pays the holder an amount proportional to the difficulty. For example, an instrument that pays difficulty/1e9 mBTC upon expiration.

Traders should be allowed to buy and naked short sell these instruments, and the price of the instrument should converge around the market's predicted next difficulty.

[fairlay]

You should create an instrument that pays the holder an amount proportional to the difficulty. For example, an instrument that pays difficulty/1e9 mBTC upon expiration.

Traders should be allowed to buy and naked short sell these instruments, and the price of the instrument should converge around the market's predicted next difficulty.

That is a future plan for our site - indeed. But the amount to loose/win would have to be bounded (otherwise you can not do it without counterparty risk). But with a range from 0-35% change really most cases should be covered.

[runeks]

You should create an instrument that pays the holder an amount proportional to the difficulty. For example, an instrument that pays difficulty/1e9 mBTC upon expiration.

Traders should be allowed to buy and naked short sell these instruments, and the price of the instrument should converge around the market's predicted next difficulty.

That is a future plan for our site - indeed. But the amount to loose/win would have to be bounded (otherwise you can not do it without counterparty risk). But with a range from 0-35% change really most cases should be covered.

Users who naked short sell should have sufficient collateral in their accounts, of course. And you would have to implement margin calls, unless you want users to post 100% collateral (which might be a problem considering that users would need to trust you with their bitcoins).

[fairlay]

Actually it turns out that the bitcoinwisdom estimation was good - the next dif. will indeed be >27B.

Finally our market agreed: https://www.fairlay.com/predict/registered/new/next-difficulty-27b/ (click on "additional stats" to see the graph)

[fairlay]

Currently we have a lot of predictions going on, at our regular "next difficulty" prediction:

https://www.fairlay.com/predict/registered/new/next-difficulty-31b/

bitcoinwisdom sees the next difficulty at 31.2B. However - if you think they are wrong or you have a different source you can currently get 7 times your money back if you predict at <31B.

Click on "show additional stats" to see the order book of predictions.