Looks like the charts, at least for https://bfxdata.com/swapstats/usd (maybe others, I haven't checked) are having some issues with the FRR.

The sidebar has the correct number listed under "FRR (% day)" but the bottom chart has the vertical line stuck at 0.05161, and has had it stuck there for some time.

I think the same stuck variable might also have been affecting the top chart, and the "Last opened Margin Funding" list. Have seen funding getting recorded in the list and on the chart at exactly 0.05161% when the actual FRR was substantially lower and so far as I can tell from the Bitfinex interface no funding was actually taken at that rate, so my suspicion is that "funding taken at FRR" was being translated into "taken at 0.05161%".

The latest changelog describes it in more detail.

A limit order that ensures the user pays the “maker” fee by only being placed if it will be added to the order book and otherwise canceled if it were to cause a match

Not sure yet whether it's a display bug or a 'bug' bug, but I currently seem to be unable to place 'margin funding' offers.

I enter the details and click the button and it says "Offer placed" but nothing shows up in the "Unfilled offers" area. Also nothing visible on bfxdata.

Oddly though, if I click "cancel all", the first time it says "All offers cancelled", but subsequent times, "No offers to cancel", as if there was something there the first time.

Edit: Scratch that, it just took a long-ass time to appear in the Unfilled Offers list.

https://bfxdata.com/swaphistory/usd

Second chart shows the average rate and the sum total of active swaps. If you zoom it out to ~1 month scale you'll see that in mid-August the total dropped abruptly from $32M to its current level of $24M and the going rate has been on a slow decline ever since because there's spare funds chasing scarce returns in a race to the bottom.

Not that the sum/rate correlation is ever that strong. Seems to be more a link from the rate of change in the sum to the going rate - not enough for funds to just be taken up, they have to be taken up quickly to move the rate up (presumably because otherwise the increased demand is just absorbed by the ever-present wall of spare funds offered at FRR).

So it's been a while, but I'm still having this problem... any time I leave bfxdata open for very long it eventually (sometimes actually quite quickly) locks up my browser for a moment before I'm asked to kill the script. When I do so it typically stays locked for another few seconds before asking me again to kill the script and then after that comes back to responding normally. If I then leave the page open the process repeats a few minutes later.

Slightly different error message though - "A script on this page may be busy, or it may have stopped responding. You can stop the script now, open the script in the debugger, or let the script continue. Script: https://ajax.googleapis.com/aj…bs/jquery/1.11.1/jquery.min.js:2 -- [Continue, Debug Script, Stop script]"

Might make more sense, but it would play hell with my attempts to keep track.

and I imagine it might also be quite a significant architectural re-shuffle on their end.

I think it has to be more frequently than once per hour - I've not seen longer than about 20mins.

As for calculating the returns... I keep unnecessarily close track of what I expect to receive, and calculating it as if the return is paid for the exact duration of the swap has been successful enough to predict the daily payment to within a few cents of the actual figure (and it isn't all in one direction so it also averages out over time)

I've seen the same thing at various times, but with no apparent connection to daylight savings and not as neat as exactly an hour.

It's the same display bug where negative numbers display as positive, but I'm pretty sure it can also happen in the space of time between when the swap was (strictly speaking) supposed to end, and when a scheduled "update/check all the swaps" batch job runs to actually close it.

FRR and fixed-rate swaps are two separate entities occupying space on the same order book even though they can't really be matched against each other (although they are commonly taken interchangeably by the auto-borrow system employed to find swaps for traders who don't do so manually. So a FRR demand can stay on the book despite the existence of lower fixed-rate offers, because those offers haven't agreed to use the variable rate (which could take the rate of return lower still during the course of the swap).

Meanwhile swaps of different durations add yet another dimension to an already complex set of offers/demands - can be the case that FRR offers don't get matched against FRR demands because the minimum duration demanded is more than the maximum duration offered.




Strange or not... when you agree to a variable rate swap, the rate varies during the duration of the swap. If you check the FAQ, that's presented as a beneficial feature - since FRR tends to rise when the price of BTC is on the up and fall when the price goes down (which is only really true in the broadest/loosest sense, but whatever), if a trader takes a FRR swap it'll (probably) only become more expensive if they're simultaneously profiting from the rising price.

Pretty sure FRR is always just one rate, for both offering and receiving funds. Or are you saying there ought to be two rates?

Problem would be that if you create a difference between the rate being paid and the rate being received, either Bitfinex would be stealing the difference (if the rate paid by traders is lower than the rate received by lenders) or there would be a shortfall in the funds collected and Bitfinex would have to pay to cover the difference and bring it up to the rate the lenders expect to receive (if the rate received is higher than the rate paid)


Fixed and variable rates aren't strictly comparable; you can't say with certainty that one is a "better rate" than the other because the variable rate could change at any time. A situation can be imagined where you would want to lock in a fixed rate that's slightly higher than the FRR at that moment, in case the FRR rises in the near future. That doesn't really explain why anyone would choose to take a 0.7% fixed rate over a 0.1% variable rate, but it does explain why the option exists to choose to not take variable-rate swaps.

I'm having real trouble figuring out what you mean here. For example, "one would create an FRR at the FRR for actively given swaps on the green side" is not a phrase that I'm able to parse.

I'm looking at the Margin Trade page, where there's a dropdown section titled "Margin available", which... doesn't seem to be entirely accurate, but attempts to say how much money is available for swaps. There's a checkbox labelled "Include variable rates?". I'm assuming that with that box ticked, FRR-swaps are included in the margin available for you to take from, and that with it unticked you'll instead only take swaps from fixed-rate offers.

When taking  a swap you can choose whether you want your demand to be matched with variable-rate swap offers or not. Which makes sense as a feature to have, because "FRR, currently 0.1%" isn't necessarily better than, say, a fixed-rate swap at 0.11%, if you expect the going rate to rise. But when someone specifies a demand for a massive amount of swap with "Don't take variable-rate swaps" checked, they can sweep out all the fixed-rate offers up to 0.75% without touching the FRR wall at all.

Possible motivation for doing so... I guess it could be an attempt to push up the flash rate. If so, it is at least a "gentler" way of moving the going rate than the old game of "It grinds slowly down for months until at 0.02% the whole FRR wall goes at once, and then the going rate jumps dramatically up to 0.75% until the wall is rebuilt"


Mostly very true - they don't last long. But if you happen to have funds repaid at the right moment it can still be worth trying to catch the spikes. I've had the rare/occasional swap last a lot longer than expected at a high rate (at least one that's still going after 2 weeks). My guess is that they're small enough quantities that the high rate passes unnoticed because the total fee remains low.

I did notice a little weirdness with the UK switching forward an hour (GMT to GMT+1) - I tried to account for the change in my records by subtracting an hour from the duration of each swap that crossed the point of the change, but the resulting prediction for how much of a swap payment I was due at 1:30 (local time when the payment was made, 00:30 GMT) was more accurate if I didn't subtract that hour, as if there had actually been an extra hour's duration in that day when the clocks changed.

Well... the negative numbers appearing as positive is only a display bug, the more salient point of negative "time remaining" being possible in the first place is at least a minor bug.

I guess they're only able to be expired when some scheduled check is run - same as the delay on re-offering funds with auto-renew.

Exactly as this guy said:


Except it seems that the server is just a little tardy when it has quite so many swaps to run that check across.

So, I seem to remember that the downloadable CSV of my account history (USD.csv) used to adjust the timezone of the timestamps according to the changes to local time; the timestamp of the daily swap payment, for example, going from "00:xx GMT" to "01:xx BST".

The clocks in Britain recently changed to BST (GMT+1) for the summer and, whilst that appears to have been reflected in the timestamps listed on the history page, it's not reflected in the CSV. Any possibility of having that back as a feature? I guess I can just manually do the "+1" where necessary, but it'd be nice to not need to.

From that initial history page, there's a row of tabs to view your history in different currencies and in more detail. The "Margin Swaps" and "Unused swaps" tabs hold your lending history. That said, it doesn't break down the payment amount any further - just lists the Start/End time, amount and rate.

Would be nice if it did break it down to a per-swap figure, but if you're worried about the payments being a bit underweight, I can (sort of) vouch for their accuracy, in that I have a long-running spreadsheet calculating the expected return (amount * rate * duration * 85%) for each swap and hence my expected return each day. Going by that, I'm fairly sure they're not cutting off any part-hours or part-days.

It's rare that I'm wrong by more than a few cents, that's very plausibly because I tend to copy the times from when the notification emails are received (which is generally a few seconds later than the real start/end time), and the total accumulated error for the whole year is about $4... and in my favour, at that.

Yes, swaps are explicitly only there to provide leverage in margin trading. What else exactly were you planning to do with the funds?

It has to be real money because it's going to go into trades with other users who might want some actual money in exchange, but it's "virtual" to prevent the swap recipient being able to run away with it. Similarly (in the sense that it's also to prevent losses) they're limited to borrowing only a multiple of the value of their deposit, and will be liquidated when necessary to pay off the swap.

The whole setup is designed to, as far as possible, guarantee that no funds will be lost by lenders. Or rather, by Bitfinex, since they've supposedly extended their "insurance" option to all swaps by default.

That looks like what would happen if someone set an auto-renew at a super-low rate as a way to always hit the highest-rate demand, and then there was no demand for swaps of that duration.

"Stop" is one of the order types in the dropdown menu. After you open your long you would enter a Stop Sell for the same amount as your long (or for less if you only want it to partially close, or for more if you want to automatically flip into a short when the Stop is triggered).

You can use the Stop order type even if you don't already have a position open - the meaning of it is "When the price reaches this trigger value, immediately do a market order", so you can use it to open a position as well as close one.

A QR code is just a blob of data in a convenient barcode form. Your phone stores that data somewhere and uses it as a seed value (some manner of hashing operation applied to the seed and the current time produces the 6 digit code), while the server holds onto the same data to independently generate the code so that it can check them.

tl;dr : yes, the QR code remains "good" forever, and can be printed on paper or stored in a screenshot (and then encrypted) for later re-use if your phone dies.

It is possible to access the memory where it's stored on your phone, to take a backup after the fact, but it involves rooting your phone and it's generally easier to just disable 2FA then re-enable it to take a copy of the new QR code.