Berkeley CIFT

Erik M.C. Peterson: If it wasn't I don't think the Bulge Brackets would be doing it. However GS proprietary trading algorithms, if revealed, could shed light on their strategies.. making them less effective. Not only that : from the article 'But no one disputes that high-frequency trading is highly profitable. The Tabb Group, a financial markets research firm, estimates that the programs will make $8 billion this year for Wall Street firms. Bernard S. Donefer, a distinguished lecturer at Baruch College and the former head of markets systems at Fidelity Investments, says profits are even higher. '

David Leinweber: HF Trading is clearly highly profitable. There are many employment lawsuits that show multi-million comp deals. The Nov issue of Advanced Trading is all HFT, and v.good http://www.advancedtrading.com/issues/200911/index.jhtml;jsessionid=AGVTEB0B1WS5BQE1GHPCKH4ATMY32JVN

Erik M.C. Peterson: side thought: perhaps E.M.H. will never be possible because of the inherent risk in the data that we are using to process price action with? For example : false news / typos - Where If I want to drive a stock upwards and I hack a few hundred drones. I have these start writing zombie blog posts and comments to 1000's of sites and boards talking about a ticker or company name.. or somehow post press releases? Trustworthiness of data will be key, and knowing what data should be excluded entirely.

Max Dama: Cool.

Erik M.C. Peterson: I call the upstairs tranche

David Leinweber: That will be difficult. It's all public information. For US publicly traded investment firms (eg Berkshire) there are SEC files w management letters You could also try asking for them.

Erik M.C. Peterson: Mutual funds usually have the prospectus, etc etc.. Are there really letters going out to investors on a quarterly basis? My 401K is handled through fidelity, and I see no such information.. except for the following: Snapshot Overview Expenses & Fees Rankings/Ratings Performance Volatility Measures Holdings Fund Facts Prices & Distributions Prospectus -------------- I don't really know what sort of value you could derive from any of these documents that you couldn't get by just looking at the HOLDING %'s

Salman Khan: Since one side effect of this plan would be New Banks lending to Old Banks, John had asked about how different this plan is relative to the the Treasury essentially picking which banks to inject capital into. This was my response: Right now, a lot of the lending is not happening (both within the financial world and to the real economy) not because of fears of insolvency, but because of the fear that the borrower won't have the ability to roll-over the loan to another party once it comes due. This explains why access to funding has become very "digital"; you either get it at a low rate or you don't get it at all. Let's say you need a $1m loan to operate a low-risk business with $200k/year pre-interest-cash-flow/EBITDA. You can clearly cover the 10% interest but I don't want to commit to you indefinitely. In a normal environment, I would loan the $1m for a 1 year term and if I didn't feel like extending/renewing it, you would find another 1m loan from someone else to pay me back a year from now. However, if I think that you might not be able to replace the loan in a year, my one-year commitment is essentially much longer (no one benefits if I take you into bankruptcy for defaulting). I have to worry about my own liquidity first, so I decide not to lend. This ends up being a self-fulfilling prophesy. On the other hand, New Banks would not be anywhere near as concerned about their liquidity so they would be willing to take the risk that you won't be able to find another loan at the end of the year. They would put covenants on the loan that allow them to get a higher interest rate if that happens, so as long as your ability to cover the interest doesn't deteriorate, the new bank will actually get an even better return. By taking this first step, the entire banking system will regain their confidence in the notion of access to funding a week, month, or year from now and would feel more comfortable in the liquidity of term loans. This and the notion that the New Banks are making a killing on "low hanging fruit" would spur the more solvent existing banks into action. If no one makes that $1m loan to you, your business will suffer and you (and your suppliers) are more likely to default on your/their other debt (further impairing the quality of assets already on "Old Bank" balance sheets. I view New Bank lending to Old Bank as a special case of the lending-to-a-business-that-is-good-for the interest-but-may-not-be-able-to-roll-over-the-loan problem, but it is not the essential argument. The essential argument is that the New Banks keep good real businesses from starving for capital and the Old Banks benefit as a side effect.

Erik M.C. Peterson: Looking back at this paper, this would have been an interesting idea. It seems the small banks are the ones paying the price though, with higher FDIC fees possibly coming in order to replenish FDIC coffers. The governments plan may have done a lot, but the failing institutions and the men/women behind these institutions are still in the system. My fear is that these people now know that the government will do whatever it takes not to have deflation, and is willing to recapitalize huge banks, but not CIT. Lesson is to get as big as possible then, and hope that you float.

lisa goldberg: you mention that an optimizer is an important tool required to get started. In a recent paper, http://papers.ssrn.com/sol3/papers.cfm?abstract_id=1074622, andrew lo and pankaj patel use the barra optimizer and risk model (and some credit suisse alpha signals) to construct long-short benchmarks for 130/30 investors. is this a worthwhile application of an optimizer, and are the benchmarks lo and patel constructed of any use?

David Leinweber: For quantitative investing, an optimizer is an important tool, in long only, market neutral, and equitized strategies. It provides a systematic means to deal with multiple risk factors. I don't see why it would be become unimportant in 130/30 strategies. The arguments for the dynamic benchmark are interesting, but likely too complex for many investment clients. Keep in mind, many public funds did not even allow shorts until recently. Some still don't. A simpler benchmark is to ask what the funds would earn if the manager made no active bets. For a long only strategies, that turns out to be an index. For MN, T-bills, for 130/30, regarding the strategy as the (weighted) sum of a long-only plus MN suggest a similar sum of benchmarks.

Max Dama: Dr. Leinweber, Realtime and historical data would be great since it's so expensive. The Haas library has access to some intra-day historical data via Wharton Research Data Services: http://wrds.wharton.upenn.edu/home/index.shtml but it's not current. Also the Haas library's subscription to the magazine Wilmott is never current. I've brought it to the attention of basically every librarian about once every 2 weeks since last spring. I'm not exactly sure what you're envisioning. Right now my "environment" is just my ultraportable laptop with Matlab and another open source statistical learning software (RapidMiner) along with data from either Yahoo finance for end-of-day or the Wharton Database for intraday. Regards, Max

Paul Hoff: You probably already know about Interactive Supercomputing and their "parallel system on demand" product which they call Star-P. You might check out their financial service page (http://www.interactivesupercomputing.com/industrysolutions/is_finance.php) for some examples of the problems that they are positioning thier solution to solve. (Also I see Star-P a pretty interesting offering which could level the playing field for smaller organisations). Julius Finance (http://juliusfinance.com/) is one of their customers. I can put you in touch with one of their board members if you want to have a discussion. This person would be intersting as he was also the founder of a start-up that made applications that exploited the parallel computer power in commodity graphics cards to run other applications. - Paul

David Leinweber: Don't feel too bad about not knowing before. The shrink-wrap is barely off the website, which only went up a week ago. BarCampBank sounds interesting. Please keep us posted on its status and dates here.

Tito Ingargiola: You should take a look at tervela (http://www.tervela.com) as they have something like a hardware implementation of CEP. Very compelling looking product though it remains out of my price range, so I can't provide a detailed product review as yet!