Topic: New Deposit Methods For TradeHill via BitInstant Partnership (Read 2985 times)

W'ere back up with some updates!

Our current rates and limits
Payment method    External exchange    You pay            Fee            You receive
MTGox Coupon    TradeHill                    $150.0000 USD    0.4667%    $149.3000 USD
MTGox Coupon    TradeHill                    $300.0000 USD    0.2889%    $299.1333 USD
Liberty Reserve    TradeHill                    $150.0000 USD    1.1333%    $148.3000 USD
Liberty Reserve    TradeHill                    $300.0000 USD    0.6111%    $298.1667 USD
Dwolla    TradeHill                            $150.0000 USD    0.6667%    $149.0000 USD
Dwolla    TradeHill                            $300.0000 USD    0.4444%    $298.6667 USD

Any amount between $20 - $500.00 USD will be accepted instantly!

Payment method    External exchange    Limit per transaction    Limit per day
MTGox Coupon    TradeHill                    $20 - $500.00 USD    $4000 USD
Liberty Reserve    TradeHill                    $20 - $500.00 USD    $4000 USD
Dwolla    TradeHill                            $20 - $500.00 USD    $4000 USD

See our Fees page for more information- https://www.bitinstant.com/fees
Quoted limits and fees are correct as of Tue Aug 30 16:51:16 2011

Thanks

Charlie

I'm in on the Ruxum beta as well, and I think they have some real potential too.  The biggest problem I see with them now is that they don't yet have any kind of API...

Withdrawals out of TradeHill are planned too, but otherwise we have an exclusive partnership deal with them which I believe has been of great benefit.

This sounds interesting, do you have more information?

Is bitinstant a tradehill only service? I think I read some statements about tradehill being the exclusive partner, or something. At the moment there seems to be not even a reverse direction, tradehill->mtgox. I thought your idea is to become the instant money "web" between all exchanges - that would mean having all directions.

Second reason I am asking is because of Ruxum. I am on the Ruxum beta. I think Ruxum will rock the bitcoin world once it goes public. They already have 12 currencies and one can now *instantly* transfer funds between currencies. In other words they have an instant currency exchange built in. So it would probably an easy add for you, because having this currency exchange, one would only need to transfer USD there and can trade all other currencies too.

If we were using a neural network on its own, i'd question things myself
Neural networks are not the magical cure-all that everyone thinks they are.

Yes, it does. I'm glad to hear that you haven't made any extraordinary claims and are upfront about the initial paucity of the training data for the neural network. I will now assume that the unfortunate quote from Pixelon's marketing material was truly accidental.

Thanks again and good luck.

I can say that I've been using this for a week or so now (since the sneak preview started) and it is pretty slick.

As for the fees, you just have to look at it like any other service.  If the fees you have to pay to use the service outweigh the profits that you can make with that money, then you don't do it.  Personally I've used it about a dozen times to get money from MtGox over to Tradehill to take advantage of better pricing there.  When you look at the other options available (really just Paxum, which is expensive and takes forever) it can be pretty competitive, especially if you need the money there immediately.

We have just updated our new fee structure, can be view here https://www.bitinstant.com/fees

Some examples:

Amount you pay   Amount you receive   Our fee

MtGox Coupons
$40 USD           $39.30000 USD           1.75000 %
$200 USD           $199.13333 USD           0.43333 %

Liberty Reserve:
$130 USD             $126.65000 USD   2.57692 %
$250 USD             $244.55000 USD   2.18000 %

Dwolla:
$60 USD               $59.35000 USD   1.08333 %
$250 USD             $249.01667 USD   0.39333 %

Let us know what you think!

Charlie

Again, in the interest of transparency, there is no "R&D team" - i'm doing the R&D solo myself, but the algorithmic trading stuff is low priority for now.
There are no published academic papers on the precise application of the methods which we are using, but the methods themselves are pretty standard.

K-means is a clustering algorithm - it takes a dataset and identifies clusters of single items of data that share similarities. The standard example is a 2D plane:

(source: mathworks.com)

In the example above, we have a 2D plane with an X and Y axis, and we have various items of data plotted on it.

Let's imagine (it's more complex than this, but just to explain) the X axis represents time of day and the Y axis represents the transaction size - plot every single transaction and you'll get some natural clusters where there's larger transactions at certain times of day. Now, imagine if scammers like to move very large amounts of money during quiet hours when nobody is around to monitor, we'd get a large cluster during the quiet hours and this would be something that can be thrown into a model.

Now, clustering does nothing in itself to help you identify fraud, and clustering in only 2D is often pretty useless - so you'd identify clusters and then look at the transactions within them to analyse how many resulted in later fraud or had other warning signs, and you'd use that to build a probabilistic model by looking at the variables in your n-dimensional space and looking at what kind of relationships there are between fraud probability and increase/decrease of particular variables.

Combine this with everyone's favourite algorithm - a backprop neural network (not as amazing as most think) and feed it to a naive bayesian classifier and other models and have each classifier output a fraud probability, weight the output of each classifier based on past accuracy and then take a simple mean average across all classifiers. You then refuse transactions that score a fraud probability above your tolerable limit based on expected losses (lookup "expected utility").
See http://citeseer.ist.psu.edu/viewdoc/summary?doi=10.1.1.37.1595 for an academic paper dealing with this approach to classification problems.

With BitInstant, the internal model does not yet have a large amount of data but over time the intention is to refine it more and more. Right now, we've identified a few factors in Dwolla transactions that make them high-risk using simple common sense and natural intelligence (i.e our brains not our computers). These variables are screened for in all Dwolla transactions as a result.

I hope that explains sufficiently, i'm sorry that I can not explain more details as beyond the above the details of the actual variables being analysed and some of the other tricks we use are trade secret territory.

Hi Gareth!

I'm sorry I wasn't clear. I'd like to see a reference to a peer-reviewed publication (in the quantitative finance or artificial intelligence fields) that describes the research upon which your R&D team is working.

The Jared Kenna's words I quoted are almost exactly from Pixelon's prospectus, except they had "fractal geometry" thrown in for the good measure.

Would you be so kind as to ask your R&D team for the actual bibliographical references and post them tomorrow or later this week?

Thank you again.

Mainly probabilistic and bayesian methods - we're slowly refining a model of what fraud looks like and what risky transactions look like using a collection of variables to uniquely identify users. Think k-means, bayesian networks and other classic machine learning methods. Nothing very unusual, but I do believe our use of these techniques is innovative.
Internally, we also have some smallscale R&D on trading algorithms to enable us to move more funds through the bitcoin network itself by buying BTC at the most optimal time possible considering all available information.

If you are happy enough without our service, then by all means continue using your current method of payment. However, if you at any point would like to try out our service with a reduced fee for volume trading then let us know. In particular, one of our specialties is in automation and inter-exchange transfers and we have some exciting features planned in this area.

We have now released a new fee structure and other misc fixes, I encourage you to check the site as we will be making constant improvements to it.

Fantastic!

Many customers have expressed interested in paying a little higher (about $0.75 total) to have their transactions done instantly.
If theres ever a need for you to get funds in fast, please feel free to use our service!

Charlie

For me, funding my Mt.Gox account with Dwolla costs me a whopping $.25 no matter what the transaction amount.  Those transaction happen fast enough for me to be happy, and not miss any market swings since I keep a fair amount in MT. Gox anyways.

In the interests of transparency, i'll tell you what inspired the more comedic style of that page - a hosting provider called nearlyfreespeech.net. While their FAQ page has a lot of casual language, it provides the information in a very easy to understand way and without being too intimidating.
Personally, I believe that adding a human touch to things is better than trying to portray ourselves in a cold and detached manner - but I apologize if the more causal tone disturbs you.

"At the time of writing"
When that FAQ was written, the highest was indeed 10%, since then we've crunched the numbers and had to update the fees so that lower volume transactions are more expensive.

Overall, this is a step in the right direction to address the need to have fast access to bitcoins.

To Tradehill and other exchanges: Apart from offering your "exchange  product", what are you doing to market bitcoins to the people out there? Are you "just" waiting for others to do PR, advertising, hoping to profit from that?

What are your bitcoin advertising / marketing plans?

Or one more step between TradeHill and Dwolla.

Dwolla caused problems in the past for TradeHill due to high fraud rates. By specialising in handling the deposits side of things, our partners at TradeHill can focus on the trading systems and not have to worry about dealing with fraud or the painful Dwolla API.
MTGox still take 15 minutes or longer to take Dwolla payments - we do it in 1 minute (well, 57 seconds to be precise).
We monitor our servers to verify performance metrics are reasonable so that we can stay to our guaranteed times and have an insane amount of automated systems to handle problems before a human need intervene.

CubedRoot, thanks for your feedback.

Yes, at the time of writing our fees were alot lower. We had a flat fee system which benefited lower volume transactions but not higher volume transactions.

At the moment, we are reviewing the fee structure and will give you an update within the hour.

Their FAQ page is also enough to make me question their seriousness as a viable business. It reads like something a high-schooler would put on their Facebook page:
https://www.bitinstant.com/faq

Bombs going off in the datacenter? Magic? Laundering huge piles of money? Come on guys, lets get serious here.

There are so many discrepencies with it, I highly question them:

From their FAQ:

But we can see on their pricing page, that they do indeed charge more than the 10% they state in their FAQ:
https://www.bitinstant.com/fees
It clearly goes as high as 20.25%

Hi, I can see how the 11.25% fee on Dwolla seems excessive, but one must consider the small volume - small volume transactions cost more to handle at these speeds (and we mean speed - under a minute is the goal for ANY payment - including the time taken to handle fraud checks which protect our ability to continue to provide the service).

The dynamic pricing system we use is intended to lower the rates as volume increases, but as I can see the high fees have been unpopular we are reviewing the fees right now and will evaluate whether the increased overall volume makes up for the lower commission.

When you use our service and pay the fee, this is what you are paying for:
 Upstream fees (Dwolla charge $0.25/transfer for example)
 Bandwidth, disk space and CPU time - we keep a full audit trail of all transactions which allow us to correct rare mistakes or errors very fast and we monitor transactions in realtime to look for problems, which leads onto.....
 Speed and customer support - we offer you guaranteed times on deposits after which we refund your fees in full
 Risk factors - our pricing system increases our rates if the current market conditions make it more difficult for us to move funds around internally (internally, the transactions can take a LOT longer than just 1 minute, we swallow that waiting period on your behalf)
 Automated payments API - for programmatic traders this is going to be huge (and we have a lot planned in this area) - sending funds in seconds from point to point B as your algorithm dictates becomes easy with our API

Currently, we are reviewing the rates internally and will release an update within the hour. Then we will review rates and transaction volume on a weekly basis. I hope this helps.