Jump to content

Earning rewards through Songbird (coming very soon)


Seoulite
 Share

Recommended Posts

1 hour ago, thinlyspread said:

Maybe, but that would make for some sweet, sweet arbitrage! 

Exactly the point I made a few days ago. When the providers of the oracle feeds are running arbitrage bots, they are inventivized to be slightly off the mean and drive the price one way, then make $$$ off trading between on ledger prices and exchange prices. The incentives could result in systematic shifts that make the system useless.

20 minutes ago, tulo said:

It can fail badly if the oracles use as input the previous price from other oracles. Otherwise if the price prediction is only a function of "real prices" everything should be fine because no positive feedback is in the loop.

nope. I can take a weighted mean of all the exchange prices, supply my feed and get a value that is just below the range of rewards by epsilon, so I add epsilon tweak to my price and next time round, I hope to be in. Once lots of providers start doing this, then can cause a bias that pushes the median/mean quoted price up or down and it can feedback to itself because everyone keeps adding more epsilons to try to get in the sweet range. Of course we hope that they all cancel out and we converge to the best true price - but I'm not certain that that will happen.

The white paper goes on about schelling games and how the price feeds are like those and therefore safe from 'bribery' - but those games assume the bribery comes from outside the system. Here, the incentives are inside the system already and if you compound it with arbitrage rewards it could get worse. Caveat - I have not done any kind of full or thorough analysis of the system.

1 hour ago, brianwalden said:

I think once the F-asset system is running, arbitrage will keep the prices in line with exchanges.

I hope you are right. Arbitrage can move the true price of the asset (closer to the FTSO quoted price) - but it can't move the FTSO quoted price directly - there is always an incentive for an arbitrage agent to supply the wrong price. (if the profits from arbitrage outweigh the loss of income from delegating - which may well happen once providers drop below the 20% "mafia cartel" figure that is in place currently). (meaning, they get more votes by dropping their prices, but gain income from bad price feeds).

Link to comment
Share on other sites

18 hours ago, brianwalden said:

Your rewards become available to claim each Saturday. This is the first week of earning rewards so the first time anyone can claim their rewards is this coming Saturday.

Seems arbitrary, is this just for SongBird (not Flare) and/or, is it just for the very beginning phase of rollout, e.g. until the entire system is functioning fully and there's more liquidity etc? 

Link to comment
Share on other sites

36 minutes ago, thinlyspread said:

Seems arbitrary, is this just for SongBird (not Flare) and/or, is it just for the very beginning phase of rollout, e.g. until the entire system is functioning fully and there's more liquidity etc? 

We're all just learning about this as we go. I haven't seen any real documentation. But it seems like the rewards run on a weekly system - they call them epochs. The rewards you accumulate during the epoch aren't made available until the end of the epoch. I think that's the system, Flare will be the same way.

Link to comment
Share on other sites

4 minutes ago, brianwalden said:

We're all just learning about this as we go. I haven't seen any real documentation. But it seems like the rewards run on a weekly system - they call them epochs. The rewards you accumulate during the epoch aren't made available until the end of the epoch. I think that's the system, Flare will be the same way.

If so, that's disappointing from my pov, from a technical perspective, given how much more advanced (in this particular area anyway) the Tezos LPoS system would seem to be, for distributing inflationary rewards every 2-3 days max without having to get in at a preset time/date (it starts *instantly* when you stake XTZ). So Flare/SongBird seems old fashioned by comparison.

Link to comment
Share on other sites

44 minutes ago, jbjnr said:

if the profits from arbitrage outweigh the loss of income from delegating

If their prices were so far off the average in an attempt to game the system wouldn’t this just make SGB holders delegate elsewhere because their rewards would be so low, decreasing that Oracles ability to influence the average and profit from arbitrage?

Link to comment
Share on other sites

1 hour ago, jbjnr said:

nope. I can take a weighted mean of all the exchange prices, supply my feed and get a value that is just below the range of rewards by epsilon, so I add epsilon tweak to my price and next time round, I hope to be in.

Adding an epsilon based on the old estimation is a useless strategy IMO. The price of an asset can be seen as white noise so addying a bias as that has the same probability of hitting the next price as adding the negative of that delta. This unless there is a history for which an oracle has constantly a bias to the "accepted" price.

BUT I'd do a different strategy which could be bad: supposing that the other oracles use exchanges' data to make the prediction I'd collect historical prices from exchanges and historical predictions, then make a model (a regression with least squares for example) of the other oracles' predictions and try to understand their future estimates. Then pick the price based on the mean of their predictions.

BTW it's an interesting problem this oracle battle :D

Link to comment
Share on other sites

The other thing isn't even collusion but laziness. How many data providers, for example are just getting to play for a ChainLink subscription and submit those prices? If too many providers do something like that, they turn into a unified block even though they weren't coordinating with each other.

I think in it will be good if exchanges who support SGB/FLR end up just submitting their price feeds as is. Ideally that raw data is what you want - actual prices from actual trades. Don't get me wrong, the modeling that data providers do to try to come up with the most accurate price is great, but the modeling is also what allows for them possibility of gamification and tweaking the model towards profitability over accuracy.

Link to comment
Share on other sites

2 minutes ago, brianwalden said:

I think in it will be good if exchanges who support SGB/FLR end up just submitting their price feeds as is. Ideally that raw data is what you want - actual prices from actual trades.

Ideally that would be great, but in the practice it's bad. Because it would be very easy to predict the price feeds of multiple oracles (the exchanges) since it's a 1:1 with their market price. Other oracles would use that information to submit predictions for profitability.

IMO the most complex and unpredictable is the price prediction of oracles the better since other oracles can't infer any information and play accordingly.

Link to comment
Share on other sites

46 minutes ago, tulo said:

Ideally that would be great, but in the practice it's bad. Because it would be very easy to predict the price feeds of multiple oracles (the exchanges) since it's a 1:1 with their market price. Other oracles would use that information to submit predictions for profitability.

IMO the most complex and unpredictable is the price prediction of oracles the better since other oracles can't infer any information and play accordingly.

But some providers are already going to do this whether or not the exchanges are FTSOs. If you can scrape the price feed from CMC and that's good enough to win some awards with hardly any effort, some people are going to find that better than doing real work to cobble together their own bespoke algorithm. Just like it's better for most people to do no work and simply delegate their vote.

It really is interesting. Once you have incentives, it becomes about game theory. It highlights how wise Ripple was to build an incentiveless system. But then that gets back to ideal vs practice. Ripple's ideal may be what's best in a mature crypto-based financial system. But crypto is still very immature. What's been hugely successful is to pay people to use your token, they use your token, and then you profit from it. Ripple has missed out on the rise of DeFi because of it.

Edited by brianwalden
Link to comment
Share on other sites

@FTSO_AU I can imagine it’s a frantically busy time for you guys.  But if you can spare a moment can you please elaborate a bit on any of the below?

  • Do you think this FTSO rewards system is likely to get to ‘true’ prices or do you agree there is the possibility of off-centre locuses of price becoming the norm?
  • Do you expect the wide rewards disparities to continue in the coming weeks/months?
  • Do you expect the returns to remain as high as they are?  Even though you are not performing as high as the leaders it’s still an excellent return.  Is that likely to continue for months and years?
  • Do you guys have metrics that help you see why there are these disparities?
  • How do you feel about the wisdom of this FTSO system versus say a UNL-like list of exchange trade feeds being simply averaged?

Anything on the above or any other matter would be exceedingly interesting to us I think.  Thanks if you can spend the time.

 

Link to comment
Share on other sites

Hey @BillyOckham ... yeh, busy times but part of that is engaging with the community as much as we can, and this is a great platform to do so. I'll keep my replies short but I don't intend to be vague.

5 hours ago, BillyOckham said:

Do you think this FTSO rewards system is likely to get to ‘true’ prices or do you agree there is the possibility of off-centre locuses of price becoming the norm?

I believe the prices will be 'true' so to speak, I don't see them deviating that far from what they *should* be. I guess we'll revisit this soon though, if / when we start to see irregularities.

5 hours ago, BillyOckham said:

Do you expect the wide rewards disparities to continue in the coming weeks/months?

I do, for so many reasons. There's currently 4 more providers ready to join the mix for the next Epoch, including Flare Finance's service which will have a large following.

New signal providers, plus new exchanges dropping tokens could see things swing around 180, it all depends on what pricing they submit and how those prices align with current providers. That's a ~20% increase in providers ... which will lessen over time.

5 hours ago, BillyOckham said:

Do you expect the returns to remain as high as they are?  Even though you are not performing as high as the leaders it’s still an excellent return.  Is that likely to continue for months and years?

They should do, for how long I'm not sure. For anyone with a nice bag willing to keep compounding and using the ecosystem I think it's a good opportunity.

5 hours ago, BillyOckham said:

Do you guys have metrics that help you see why there are these disparities?

We're reporting internally on what the successful price submissions were and what prices we submitted. We're using that to better understand our position in it all. There's not a lot in the numbers either, it doesn't take much to miss the median, either high or low.

5 hours ago, BillyOckham said:

How do you feel about the wisdom of this FTSO system versus say a UNL-like list of exchange trade feeds being simply averaged?

It's all so new, despite everyone reading about it for the last however long .... we're only really truly seeing how it behaves now we're finally into it. 

5 hours ago, BillyOckham said:

Anything on the above or any other matter would be exceedingly interesting to us I think.  Thanks if you can spend the time.

Like you, we're pretty pumped to be past Observation mode, that would have been a real kick in the guts.

As I said in my opening line, this particular community is awesome, always lots of great conversation. It seriously wears you down with the "wen launch" "how much will I earn" noise elsewhere, so getting to where we are today and being able to move past that is a big relief. 

Don't get me wrong, we're not just here for the good times, I'm happy to hear the truth too ... and we're certainly underperforming on the "winning price rounds" aspect on what we'd hoped, but any changes we make will be done so with good reason. All other technical aspects of our service is working great, which is really pleasing!

Link to comment
Share on other sites

21 hours ago, FTSO_AU said:

We’ll, it actually just ticked over 48hrs, or 2 days, so we’re a long way short of a week, but that’s irrelevant.

Saturday night, 6pm my time was the first time the FTSO was tested in full battle conditions, up until then it was only prices being sent and revealed every 4mins, which we’ve done successfully, and continue to do so.

The rewards started off promising, but drifted south 24hrs later. We’ve analyzed a lot of data and deployed an algorithm change midnight Sunday, after several hours in the test environment. 

The results didn’t head in the direction we’d hoped. Interestingly most other providers rewards also slowed in this period.

Todays been another day of developing some tools to further analyze the data and plan the next move. 

You can expect to see lots of swings in the reward distribution over the coming days and weeks as all providers adjust their services.

If people feel moving to another provider is the best solution for them, then so be it. That will give us the opportunity to see how the vote power cap affects the rewards when they’re diluted among the over subscribed pool.

Its not a sprint, it’s a marathon … and this is exactly what Songbird was designed for, testing our services before being deployed on Flare mainnet.

I'm picturing a horse race in my mind. All the ftso providers are the horses, in this thread are all the gamblers and FTSO_AU is the telecom announcer giving us updates. 

Link to comment
Share on other sites

On 9/26/2021 at 3:40 AM, BillyOckham said:

I don’t understand how all you folk are able to know what providers are doing well and poorly.

 

I only have two and their percentage is unbalanced so without a bit of maths I can’t be sure which is better.  But even if I do the math that only tells me my two.  Doesn’t tell me if they are better or worse than others.

How do you guys know all these relative performances?

 

https://flaremetrics.io/ftso

Link to comment
Share on other sites

5 hours ago, BillyOckham said:

Do you expect the returns to remain as high as they are?  Even though you are not performing as high as the leaders it’s still an excellent return.  Is that likely to continue for months and years?

Of course not. There are only 800M SGB delegated. As soon as more people will join and more tokens are available in the market (airdrop from exchanges) the reward will drop proportionally. I'd expect at least 2-3B SGB delegated, which means around 50% APY.

Link to comment
Share on other sites

Create an account or sign in to comment

You need to be a member in order to leave a comment

Create an account

Sign up for a new account in our community. It's easy!

Register a new account

Sign in

Already have an account? Sign in here.

Sign In Now
 Share


×
×
  • Create New...

Important Information

We have placed cookies on your device to help make this website better. You can adjust your cookie settings, otherwise we'll assume you're okay to continue.