Estimated Time to Win Explained

Who knew that this metric could cause so much turmoil. I’ve read many many frustrations from members of the Chia community that they are well past the Estimated Time to Win (ETW) and they have not won any Chia Coin. In one sense, their frustrations are rightly justified. Its justified because this metric is, in my opinion, incorrectly name under the current environment (no pools). I can see where the Chia Team were going with it, and they have made some updates to the GUI to help explain it better. It used to be called “Expected Time to Win” but they have made the change to “Estimated”. Also, the tool tip now tells you that it could be “2 to 4 times this estimate”. These are steps in the right direction. But, I think this metric should have been called “Chance to Win Daily” in the current environment. Note: This is a working document that I hope to update in the future.

Just like in casino games, you can see the “Odds” of any game type; 1 in 30, 1 in 158, etc. This is essentially the same environment that Chia is operating under. To understand this better, here is a rough overview on how Chia works. On the Chia network, there are roughly 4608 chances to win a block per day (bottom of pg. 9). I say roughly because its a moving target as new users come online and netspace grows. Your chances of winning daily is proportional to how much plotted space you have versus the netspace. Knowing this we can easily come up with a formula to figure out your chance to win. (Edit: I made a grave math error here and corrected it, Thanks to Freeze at the ChiaForum for pointing it out!)

1-((1-<YourSize>/<NetSpace>)^4608)=Daily Win Chance One Block

Let’s do the experiment. You are farming 90 Tebibytes (TiB) of plots. And for simplicity, lets say the NetSpace is 600 PebiBytes (PiB) (Side note: This is a mind boggling number. The netspace is exploding). Crunching the numbers results in a daily chance of 49.91% to win at least one block. These look like good odds! With the current way this chance is being displayed the ETW would say “in a day” because its close to 50% for one day. This is where we get to the misleading part. With the way its worded, its very easy to fall into the gambler’s fallacy:

a failure to recognize the independence of chance events, leading to the mistaken belief that one can predict the outcome of a chance event on the basis of the outcomes of past chance events. For example, a person might think that the more often a tossed coin comes up heads, the more likely it is to come up tails in subsequent tosses, although each coin toss is independent of any other and the true probability of the outcome of any toss is still just .5.

Definition from the American Psychological Association. To interpret it to our community, “I have gone X days without a win, surly I am due for one!” There were many frustrated people when it was “Expected Time to Win” because the term “Expected” infers that you are owed the win. Now with the term “Estimated” it is a little better but when you miss the time window you still feel like you are owed something. This is why I think “Chance to Win” is better and maybe allowing the user to cycle the percentages to per week/month. This does infer that this is gambling though (which it is), which could result in negative PR and currently, PR is important for global adoption. If it said “Chance to win” people would automatically know, “Oh, I have a 50% chance of winning one block each day”. When they go days without winning, they will say “I’ve been unlucky.” Instead of, “Where are my Chia Coins? It says ‘a day’!”

To continue our example, for simplicity lets say 50% daily chance to win, what happens if there are 5 days in a row with no wins? What are the chances of that happening? Since the chances are independent, we will need to enter the world of probability. Here is how we can figure it out. Take the chance per day and multiply each day that chance occurred. 50% can be written as 1/2.

 1/2 x 1/2 x 1/2 x 1/2 x 1/2 = 1/32 = 0.03125 = 3.125%

This seems like a low chance of happening, but, it is easy to find some people that have gone days without a win in the Keybase chat and online with a ETW of “a day” or less. What gives? Well, the unique address count that has won at least one XCH is almost 7000 as of this writing. That would result in 218.5 people getting a 5 day “dry spell”. But there are flaws here because not all 7000 have the ETW of “a day”. So it is definitely less but now I can see how there can be a group of people with a dryspell. The opposite is also true here, there will be the same percentage of people that win every day for 5 days in a row.

Lastly there is an additional system underlying the reward schedule that you should know about. Chia Rewards come when you have the answer to a challenge, a “proof”. The network keeps the 4608 wins a day in check with a variable called “Difficulty”. You can see this on your Full Node Tab. When NetSpace grows rapidly, Difficulty Skips ahead instead of incrementing ahead. You can see this with Chia Explorer’s graph:

Each bump up means that its harder, and harder to find a “proof”. You can read more for yourself in their consensus document.

Update (4/25): To make it clear, understanding Time to Win does not mean that nothing is wrong with your Full Node either. If you’re in the statistically impossible area (more than 4 times out of your ETW) you must check that everything is running ok. Here is a list of what to look for:

• Sync issues – If you are having sync issues, you need to resolve these first. This can happen if you:
  • Run multiple PCs with the GUI on the same LAN.
  • Having UPNP disabled on your router, but are not forwarding port 8444.
  • Having no more than 10 peer connections on the “Full Node” tab in the GUI.
  • Not seeing “Last Attempted Proof” challenges in the GUI come in about every 10 seconds.
  • Look at my post here to try and resolve some of these issues.
• Ensure proofs are returning in under 30 seconds– There is a 30 second time window to respond to a network challenge. If your machine or harvester takes longer than 30 seconds, you will miss coins. Check your chia logs and look for “eligible” then scroll to the end of the line. There is a timestamp here, in seconds. This must be below 30 seconds. A member of the community, codinghorror, is dealing with this at the moment over at the ChiaForum. If you don’t know how to look at the logs, look at my post here
  • Also, If you’re running a farmer-harvester setup, ensure that the harvesters can talk to the main farmer in under 30 seconds.