Algo Trading

[lionfish42]

Quote from rosy2:

i think you dont understand what these are used for. They are for getting into(out of) large positions. What good are they if you put on a max position of 1000 shares. How do you think they will help you?

I understand that if you are trading 100 shares that most things are not very useful or if you are a day-trader. However, I am neither - if you read my earlier post.

However, it doesn't seems that all of them are for as you say "getting into (out of) large positions." Are we not all "getting into (out of) positions". Does the scale make THAT much of a difference as you imply that it does?

For instance what are "Implementation Shortfall, Rapid Dark, Synthetic Peg" just to name a few.

I remember over 10 years ago that Level 2 and Mark Depth and other similar things were for Market Makers and they were said to be "Useless" as well. Now everyone uses them.

It sounds like, not trying to take a jab, is that you are making more of an assumption about them than actually being in the know.

 

[lionfish42]

Quote from USAtrader:

I don't much use them, but these are not highly sophisticated algo's only available through select providers. Even IB offers a bunch as standard execution algo's.

I am not concerned if they are highly sophisticated or not - I am interested in what they do and how they work. How many people use them and have they benefited there trading.

Some of these do not seem "Standard" either.

Maybe the trading community is just very slow (as it always seems) to adopt new technology and systems. I guess that is always the case.

Few early adopters and the vast majority sticks with what they know.

 

[Pippi436]

Quote from lionfish42:

I understand that if you are trading 100 shares that most things are not very useful or if you are a day-trader. However, I am neither - if you read my earlier post.

However, it doesn't seems that all of them are for as you say "getting into (out of) large positions." Are we not all "getting into (out of) positions". Does the scale make THAT much of a difference as you imply that it does?

For instance what are "Implementation Shortfall, Rapid Dark, Synthetic Peg" just to name a few.

Most of those are infact designed to minimze market-impact and information-leakage for larger orders. Historically they have been developed to minimze costs for the sell-side - its just much cheaper than a hands-on trader - but most likely perform worse than an experienced execution trader. On the other hand, its much easier to track how efficient an execution-algo works - i.e. how well a VWAP-algo tracks actual VWAP, whereas its as easy to evaluate performance of a human trader.

Implementation Shortfall: read the book i linked, its not explained in 3 lines. IS is a way to measure transaction-cost, and obviously these algos try to minimize the same. Its not for daytrading either. Synthetic peg is self-explanatory. "Rapid Dark"? I dont know, probably one out of countless algos that tries to hit dark-pools or anonymous crossing networks first before it routes to exchanges/ecns.

 

[lionfish42]

Quote from Pippi436:

Most of those are infact designed to minimze market-impact and information-leakage for larger orders. Historically they have been developed to minimze costs for the sell-side - its just much cheaper than a hands-on trader - but most likely perform worse than an experienced execution trader. On the other hand, its much easier to track how efficient an execution-algo works - i.e. how well a VWAP-algo tracks actual VWAP, whereas its as easy to evaluate performance of a human trader.

Implementation Shortfall: read the book i linked, its not explained in 3 lines. IS is a way to measure transaction-cost, and obviously these algos try to minimize the same. Its not for daytrading either. Synthetic peg is self-explanatory. "Rapid Dark"? I dont know, probably one out of countless algos that tries to hit dark-pools or anonymous crossing networks first before it routes to exchanges/ecns.

Thanks very much.

I (we) trade from a long or short portfolio position and use options a hedging and income generation. So volume can sometimes be larger than what a typical day-trader would be accustomed to.

Since our auto-mated system has pricing sensitivities to the net long/short equity position we use market price and volume to determine when to purchase/sell shares. Right now we use our own algo system, but are looking into alternatives.

I realize that probably most members on these forums are day-traders, they care little about pricing improvement and more about the "action". I would think however, that there are probably some algo-systems that would benefit even their very-short-term tick type day trading. Of course I could be wrong.

Maybe these forms are not the right place to ask about systems that are geared for a different type of trading/investing?

Thanks again - BTW I did order that book.

 

[Pippi436]

Hi,

i read the book with a similar expectation, interested in microstructure and things like, for example, to determine when its better to place a passive limit and when its better to agressively cross the spread - to archive a possible price-improvement on a very short-term horizon.

There is very little information about his kind of stuff in the book. Almost all of the 'known' standard algos that are offered by banks or brokers work on a larger timeframe, atleast those described in the book.

I guess the stuff you are looking for already touches the marketmaking/HFT area, and noone is willing to share their secret sauce in public. Someone correct me if im wrong, but im not aware of a broker offering such algos, except for smart-routing and similar.

 

[asiaprop]

as someone pointed out large buyside firms (Fidelity & Co) use them to trade in and out of positions. They constantly panic to be cheated by their execution brokers (front running, mispricing, hiding large orders a broker has knowledge of,...). Generally there are two groups of VWAPS, guaranteed vwap, where the client is guaranteed to pay/receive vwap and the risk is taken by the sell-side firm. Such algos generally demand 15-25 basis points in commissions. Then there are best-effort VWAPS, where the algorithm attempts to match/beat market vwap but the risk still lies with the client. Such algos can only demand commission in the region of a few single basis points, sometimes down to 1-2 bps.

Some better vwap strategies look at dynamic volume during the trading day rather than stubbornly using historical volume profiles to makes assumptions in what time bucket volume trades. If you google a little further you will see that at this exact point terms such as "implementation shortfall" and "volume in line" "volume participation" come in.

Some even more sophisticated guaranteed vwap strategies take additional risk by identifying price points "with edge" and take additional risk on top of pure volume matching. Sell-side firms can do that because they run the risk of guaranteed vwaps anyway.

Hope this helps.

Quote from lionfish42:

I am not concerned if they are highly sophisticated or not - I am interested in what they do and how they work. How many people use them and have they benefited there trading.

Some of these do not seem "Standard" either.

Maybe the trading community is just very slow (as it always seems) to adopt new technology and systems. I guess that is always the case.

Few early adopters and the vast majority sticks with what they know.

 

[lionfish42]

Quote from asiaprop:

as someone pointed out large buyside firms (Fidelity & Co) use them to trade in and out of positions. They constantly panic to be cheated by their execution brokers (front running, mispricing, hiding large orders a broker has knowledge of,...). Generally there are two groups of VWAPS, guaranteed vwap, where the client is guaranteed to pay/receive vwap and the risk is taken by the sell-side firm. Such algos generally demand 15-25 basis points in commissions. Then there are best-effort VWAPS, where the algorithm attempts to match/beat market vwap but the risk still lies with the client. Such algos can only demand commission in the region of a few single basis points, sometimes down to 1-2 bps.

Some better vwap strategies look at dynamic volume during the trading day rather than stubbornly using historical volume profiles to makes assumptions in what time bucket volume trades. If you google a little further you will see that at this exact point terms such as "implementation shortfall" and "volume in line" "volume participation" come in.

Some even more sophisticated guaranteed vwap strategies take additional risk by identifying price points "with edge" and take additional risk on top of pure volume matching. Sell-side firms can do that because they run the risk of guaranteed vwaps anyway.

Hope this helps.

Thanks for some details.

I am going to run a series of live test with one of our accounts using these types of algo trades and run a detailed comparison against DMA routed orders. It would be interesting to see if and how much improvement.

A couple of algos also were set up for the liquidity rebates as well. Which would be interesting vs. a price improvement analysis.

Since these algos are made to non-institutional clients like myself, why not take them for a spin at the very least.

Obsidian offers them for all clients (retail and firm). This limited information is available in their module section under Algorithmic Order Ticket: http://www.silexx.com/obsidian/modules

 

[GoldStandard]

As others have said, these algos are mainly to buy/sell large positions without moving the market against you & broadcasting your intentions. If your positions are large enough that when you buy or sell , you move the market against you, then these algos might benefit you. Otherwise there is not much, if any benefit for you in using them.

Unlike level 2/market depth these algos don't provide you with any additional information. They're not more 'advanced' , they're just tailored to the needs of particular (large) traders.

 

[asiaprop]

other algos may reduce market impact on large orders, but vwaps main purpose is actually not to reduce market impact of large orders (a lot of smaller orders that are far below 20% of ADV are traded as vwaps as well) but because buy side execution traders are benchmarked against vwap. For example, there is no upside for an execution trader to outright buy a position when his portfolio manager instructs him to buy into a position at a daily average price or vwap price (possibly with upper limit).

But I agree there are other algos that attempt to inject size into the market at opportunistic bid/offer levels.

Quote from GoldStandard:

As others have said, these algos are mainly to buy/sell large positions without moving the market against you & broadcasting your intentions. If your positions are large enough that when you buy or sell , you move the market against you, then these algos might benefit you. Otherwise there is not much, if any benefit for you in using them.

Unlike level 2/market depth these algos don't provide you with any additional information. They're not more 'advanced' , they're just tailored to the needs of particular (large) traders.