Talk:Program trading

Definition is incorrect
I agree with the comments below. —Preceding unsigned comment added by 71.238.70.93 (talk) 03:33, 15 October 2008 (UTC)

Yep. This article is all wrong. Jive Dadson (talk) 01:29, 20 April 2009 (UTC)

"Pricing stocks in penny increments instead of 1/16 increments results in 100 price points within a dollar instead of the previous eight price points. That means all the willing buyers and sellers are dispersed over many more prices, making it more difficult for them to meet on price." This phrase in the intro is illogical. Supply and demand intersect where the market clears. I don't understand how the subdivisions matter. In terms of microstructure, the smaller divisions should help, correct? —Preceding unsigned comment added by Richardcherron (talk • contribs) 15:13, 26 July 2010 (UTC)

Problems with this article
This description of program trading in this article is entirely inaccurate. Program trading is the execution of large baskets of (typically) equities, with the aim of minimising "market impact", which is the propensity for a large trade to adversely move the price against itself before execution is complete. Program trading is a service offered by investment banks, usually to large investment funds such as pensions, index trackers and the like.

The trade can be executed by the investment bank in an "agency" fashion, or "as principal". As agency, the client bears the price risk of execution (if they are filled at 39.5, they will receive 39.5). When trading agency, the bank will often/usually attempt to achieve the VWAP (volume weighted average price) daily benchmark for each stock. Typically such execution will be done by "slicing" the order into smaller orders, in an attempt to reduce the market impact of the trade. This slicing can be done linearly - 10,000 shares gets split into 10 slices of 1000 - or may follow the volume curve, where trading is heavier at the start and end of the day and the slices are weighted accordingly. When buying, a price above VWAP means they have over-performed, whereas a price below indicates under-performance. The opposite naturally applies when selling.

As principal, the investment bank will "purchase" the program from the client without knowing its contents, and will then aim to execute it at a better price than they paid the client for it. The latter approach obviously involves larger risk for the investment bank.

Algorithmic trading *may* be considered a subset of program trading, but only in the sense of algorithms that attempt to achieve a benchmark price.

Program trading typically caters to customers who "must" trade - they cannot elect to keep an order out of the market, even if it will adversely affect their price, because they are bound by some condition such as always holding a certain percentage of US equities, etc., etc.

I am happy to dig out some references and things for what I've written above (I've been working in PT for nearly ten years), and turn it into a proper article.

Program trades need to be specifically marked as such when submitted to the exchanges, and there are certain restrictions placed on programs that do not apply to non-program trades (NYSE rule 80-A, for example).

what rule?

... and at the same time to trade slowly enough so that you do not impact the stock by "walking the book".

walking the book?

S Sepp 12:30, 13 June 2006 (UTC)

Also, this article seems rather US centric - there seems to be a lot of it going on in Europe too. —Preceding unsigned comment added by Laos (talk • contribs) 21:12, 22 May 2011 (UTC)

walking the book
Rudimentary program trading began in the seventies, with the trades in the program being walked around to the market maker's (specialist's) posts at the New York Stock Exchange.http://www.econlib.org/library/enc/ProgramTrading.html

"Program trading is extremely popular in hedge funds" isn't entirely true. I think "Program Trading" is most popular with institutional investors who manage large funds. Hedge funds often try to profit off from program trades.


 * Goldman Sachs is by far the largest program trader, as big as the next largest 14 combined. Jive Dadson (talk) 01:29, 20 April 2009 (UTC)

The last line "Computers also allow traders to test their strategies against historical data in an attempt to predict and optimize them for future conditions." has nothing to due with the article.

testing strategies
Give the last line its due. Maybe it means "The program trading technique also allows traders to test their program trading algorithms against historical data in an attempt to predict and optimize their program trading algorithms for future conditions." —Preceding unsigned comment added by 218.132.197.145 (talk) 03:14, 24 May 2008 (UTC)

September 2009; copyvio
Folks, I'm sorry if I was not clear in my first reversion of the recently added material, but until someone can clear up the apparent copyvio from this site, the addition will be quickly removed. I would encourage editors to use this talk page before making any further wholesale reverts. Kuru talk  01:19, 3 September 2009 (UTC)

Whole article needs lots of work
There is no longer a 1,000,000 share requirement according to the latest NYSE releases. Additionally, there was a cite discussing program trading including portfolio insurance etc in an earlier version. I will find it and add it back. Also, program trading and algorithmic trading are not one and the same. You can do one without the other. for example.Sposer (talk) 01:21, 3 September 2009 (UTC)

Still Inaccuracies, which I will fix
(1) NYSE definition says nothing about $1mn dollars (2) Inline references are messy (3) The lede makes it sound as if index arb is program trading. I will clean it up. (4) No need for spam on HL Camp. Who says they are the only reliable source? (5) Statement about GSCO profits only being program trading has no source. FT article doesn't say that. (6) Program trading is properly measured as volume executed in programs divided by buys and sells, so that percentages quoted from the NYSE historical PDF need to be halved.

If I see anything else, I will not in the edit summary.Sposer (talk) 17:19, 3 September 2009 (UTC)


 * Just wanted to explain the other edits I made:

(1) There was a statement that said that program trading (PT) and high frequency trading (HFT)and algorithmic trading were one and the same. This is not true. Although some PT might be HFT, not all PT is HFT. And, HFT employs many many strategies that are not PT as defined by this article. Yes, HFT uses computers, but the definition of program trading, is not trading using a computer. Similarly, algo trading is not always HFT, although it might be. And algo trading isn't always PT. There are certainly overlaps, but they are not the same.

(2) I altered some text, mostly for clarity, and hope I kept the meaning the same. I changed the bit where it said the S&P 500 contract was a "commodity" contract. It is more correct to say futures, since we usually refer to commodity contracts regarding agricultural and metals futures. The S&P 500 futures would be a financial futures contract.

(3) In-line references are ugly, so I put them into references. All the information is there, and that better matches the standard Wiki formatting.

(4) The external links almost all repeated the references, so they were removed.

(5) The percent of trading that is program is more accurately depicted as the percent of volume divided by the sum of buys and sells. This is more commonly known as TTV, or twice total volume. That is how the NYSE reports it and is why I cut the numbers in half. Say there is a program that includes the purchase of 100 shares of NYX stock. Using the higher number, we would say that PT was 100% of that trade. However, this is an inaccurate depiction. Somebody else sold that stock. If it wasn't a program trade, we can safely say that 50% of that trade's volume was PT. In theory, using the old methodology could result in an impossible result, which would have been that PT was more than 100% of trading, which is of course impossible. So, the proper calculation is volume from all program trades (whether purchases or sales) divided by the sum of all buys and sells (i.e., TTV).Sposer (talk) 12:39, 4 September 2009 (UTC)

This message is for Sposer:

Sposer, that article from Investopedia cites the HL Camp website as the best and only public source for Program Trading buy sell levels. So its not an opinion why I am adding that to the article. It is something verifiable by a very respected source. Whether its a sales site or not, they provide the numbers being referenced, and according to Investopedia and the author of that article they are the only public source for that information. So it is warranted to include them on this article. —Preceding unsigned comment added by Razeboc (talk • contribs) 15:28, 4 September 2009 (UTC)


 * Wiki doesn't link commercial sites in general, nor does it typically source other Internet encyclopedias as far as I know. You need to go through multiple links to get to the program trading buy and sell levels, and need to read two paragraphs from HL Camp on why their numbers are right, but might be different from what other companies say they are. They also say that their numbers might not be what other firms are doing PT at, and that they are using their own proprietary model. None of that would be considered RS for Wiki purposes.Sposer (talk) 15:39, 4 September 2009 (UTC)

Program trading = basket trading or = algorithmic trading?
For the European guy I am, the article and above discussion are confusing. If program trading = basket trading, I suggest the article be renamed "basket trading". But in Europe, program trading and algorithmic trading are the same. I would then suggest merging this article with algorithmic trading. In any case, the article should not be restricted to NYSE's definition. Instead there could be a paragraph "NYSE's definition of program trading", and another investigating about other non-US regulators' own definitions, so that the overall article could look like more universal. Bmathis (talk) 11:40, 21 February 2010 (UTC)
 * There are multiple types of program trading. I don't think the definition just the NYSE definition, but I am not sure. A program is not necessarily algorithmic, but I do agree that the article needs work.Sposer (talk) 01:20, 22 February 2010 (UTC)

Contradiction
In the first paragraph, the article states that: As of July 2012, program trading made up about 25% of the volume on the NYSE; index arbitrage made up less than 1%. In a subsequent paragraph, index arbitrage is described as a particular type of Program Trading [sic]. Aboctok (talk) 08:48, 3 November 2023 (UTC)