Reputation
Next tag badge:
73/100 score
24/20 answers
Badges
1 7 32
Impact
~284k people reached

  • 0 posts edited
  • 21 helpful flags
  • 245 votes cast
Sep
19
comment Robust Returns-Based Style Analysis
I do not see why a risk-adjusted return analysis (which Sharpe is) is flawed just because a fund frequently changes strategies. Sharpe calculations have nothing to do with linear regressions. Of course there are improvements that can be made, I am not saying that Sharpe measures are perfect, but it still is industry standard by which you, funds, anyone is measured before people take a closer look.
Sep
18
comment Calculating arbitrage- S&P 500 stocks vs S&P 500 Index future?
What I listed are key concerns that determine whether the arbitrage makes or loses money, they are far from being operational. Your answer links to the no arbitrage argument to price a future/forward.
Sep
17
comment Calculating arbitrage- S&P 500 stocks vs S&P 500 Index future?
this has very little to nothing to do with the question at hand. The challenges arising from index arb are the necessity to execute multiple orders at very low latencies and to overcome the bid-offer spread in the underlyings, traded, the probability of getting filled when submitting orders around mid levels (its almost a given that index arb turns out to be a negative expected value proposition when being forced to be a complete market-taker), and the challenges to remain fully hedged when trading the arb.
Sep
17
comment Pre-Trade Slippage Costs For Option Spread Execution
you are omitting the most crucial piece, which is queue priority and how aggressive/defensive market makers are.
Sep
16
comment Time series analysis on illiquid price data?
what makes you think it is initially mispriced? And towards which theoretical mean, given there is no reference level to gravitate towards? As said, using other asset's time series is possibly one of the worst proxies you could pick. But that is just my 2 cents without exhaustive research on the topic. Why don't you test on your theory on past IPOs?
Sep
16
comment Time series analysis on illiquid price data?
I would strongly discourage you to estimate the first trading price of A as function of current (and even historical time series) of comparable companies. The opening trading price is best estimated from A's own fundamentals as well as subscription price, support period price bands, which banks underwrote the stock, how much the company intents to float, public interest, .... You are in for a tough exercise with tons of margin for error if you attempt to price the trading range based on where comparables trade(d)
Sep
12
comment Long/Short portfolio return
This question is answered several times in other questions on this site. Please search for "compounding returns", "aggregate returns"
Sep
11
comment Index arbitrage with Options when not all underlyings have options listed?
very good points. +1. Clearly "arbitrage" might as well be voted one of the most misused terms in today's finance.
Sep
11
comment Index arbitrage with Options when not all underlyings have options listed?
What are you attempting to arbitrage here? Your setup would only make sense if you look to extract alpha through trading implied volatility. Can you be more specific in order to understand what you are attempting to do? "Arbitrage" is nowadays such a misnomer describing anything and everything while in actuality arbitrage has a very closely defined meaning.
Sep
11
comment Should I use QuickFix or a 3rd-party commercial API to connect to the CME
...which just means that the exchange, like most other vendors that host fix connectivity, want to ensure a minimum safety check framework is maintained. I fail to understand why you relate this to the FIX engine. Certification is about testing FIX connectivity, order submission, handling, and fill management and it applies to whatever FIX engine you end up using. Obviously some proprietary FIX engines help with those tests and some open-source engines are less feature rich.
Sep
10
comment Should I use QuickFix or a 3rd-party commercial API to connect to the CME
Why are we talking certification when you just told me that you do not need to deal with it?
Sep
10
comment Should I use QuickFix or a 3rd-party commercial API to connect to the CME
As long as you do not mind about additional latency in the double digit millisecond realm you can safely go with QuickFix. I actually also use QuickFix for non latency sensitive order submission and it works fine despite a number bizarre design issues (such as MessageCracker,...)
Sep
10
comment Should I use QuickFix or a 3rd-party commercial API to connect to the CME
It completely depends on your use case. Do you require certification at the company you work at? Also, QuickFix requires quite an amount of additional code, most proprietary FIX engines already provide solutions to connect with most brokers, ECNs, and Exchanges. Then there is of course the question what latencies are acceptable to you. Quickfix is not the fastest out there, forget hft using Quickfix.
Sep
1
comment Monte Carlo Options Probability Calculation
That was my point, there are ultimately unlimited ways how to model stock prices, some of which more standard than others. I am afraid your question is too broad and to answer it properly requires writing a whole book. Hence my suggestion to search for MC-related posts on this site, play with some of those suggestions and come back and ask more targeted questions (search for "Monte Carlo stocks").
Sep
1
comment For a interdays trading backtest system, should I put day open, close, high, low, volume separately into array?
I second your latter paragraph. In some cases storing the data in segregated arrays will be computationally more intensive than storing data of identical time stamp in a struct or class. I coded up my own binary data store and query engine and it ended up to be a lot faster than kdb for standard queries. The data store manages blobs by time stamp rather than segregating out by columns. Obviously my own custom solution only supports what I need and a fraction of what kdb offers. It all comes down what queries need to be supported.
Sep
1
comment Monte Carlo Options Probability Calculation
BS: Simpler computationally speaking but very limited in regards to payoff functions, most non vanilla options do not have closed form solutions. MC: More computationally intensive but very flexible in its usage, including your choice of underlying price driven process. But this is not a forum to walk people through the basics of option pricing or Monte Carlo applications. Please google those topics for solid introductions and/or search this forum because questions about MC already exist here.
Aug
29
comment VSTOXX Implied Volatility Calculation
thanks for additional comments, I still feel 1.4-1.6 is extraordinarily high even for near expiries but not saying its impossible.
Aug
22
comment Black (1976) model: boundary conditions with non-convergence of spot and forward prices
@JoaoSerafim, I deleted my comments because I felt they were misleading. Quick summary of what I have been trying to say: The Black model is just the base framework. Depending on the underlying-forward/futures relationship (equity futures, commodity futures, currency futures, index f), the Black framework needs to be slightly adjusted and you end up with an extension of the Black framework, such as with Garman-Kohlhagen.
Aug
22
comment VSTOXX Implied Volatility Calculation
thanks for the update. Could you please comment on the y-axis units? It says ImpliedVol but the values do not correspond to reasonable implied volatility levels.
Aug
21
comment VSTOXX Implied Volatility Calculation
@Eli, would you mind sharing some of your findings? Of course only material you do not define as edge, I am simply curious what kind of analysis you have performed and the results you found as far as you feel comfortable commenting on. Would be a rare but highly pleasant treat.