Projects

A successful FX prop. trader seeking an alternative execution venue commissioned me to build a pricing repository and full trading capability to SaxoBank.  @Hwalaz

Responsible for supervising trade flow and production systems at start-up hedge fund Ashdenden.  Continuously trading global markets from multiple strategies. I built systems to automate and monitor including the PMS from the ground up, and handled and traded corporate actions, futures rolls, 3 way reconciliations, vendor relations, and a lot more. Also at hedge fund Marshall Wace I managed the trade-ops between various models and the firm-wide services. @Ashenden @MW

Once the models are running live there is the need for continuous trading operations. Handling activities such as model data validations, target portfolio validations, fund flows, trade netting, model scheduling, reconciliations, restrictions.

At MW I also built model position management systems (pre-trade) which would be reconciled to the firms system.  At Ashenden I built the post-trade position management and recon systems.  @Ashenden @MW

Onboarding models into a production trading framework.  I've played a key role in building these services at Marshall Wace and again at Ashenden for our alpha capture model.  This was done in such a way that the quant could just add a few calls to the model to provide both sim and production framework interfaces.  @MW @Ashenden

When and what to roll...   need to codify the futures contract roll rules, build a schedule, and then overlay the liquidity in near, far, strips...  and apply past data/experience based on how fast that liquidity transitions over time for each market.  Decide on whether to cash settle, configure to trade out the near and into the far...  reallocate to strategies.  Some markets/brokers can't trade a calendar-spread...  @Ashenden

Building (virtual) positions and capturing the P&L of each trade, position, and contributor.  Adjusting for corporate actions.  Handling trade corrections, and then also doing the mark to market at different benchmarks.  I also built a realtime view ranking contributor P&L (broker summary) with switchable prices between Reuters and Bloomberg.  Precursor to TipsMining.  @MW

@everywhere

Touch, VWAP15, VWAPDay ...  I built this as an overlay to the MW TOPS contributor real-time P&L system.  And then again as a standalone system later.  @MW

What started as a project to gain better insights into systematic trend-following and Python development became quite an endeavour.  Leveraging publicly available research (thank you AHL) and learnings on the job at Ashenden (start-up systematic hedge fund where I led trading ops and execution), I built an excel sim and then evolved this into a python/pandas framework utilising realtime pricing, multiple markets, simple signals (e.g. ema with stops for various look-backs),  risk management (vol normalised targeting), backtesting and calibration.  @Ashenden, @Home

I built the PMS from the ground-up at start-up hedge fund Ashenden.  

Born from the need to have a timely view of trade flow, aggregations over strategies, reliable reconciled positions and adjusted history...  The PMS formed the backbone for active trading and post-trade services and sourced directly from the FIX feeds (> orders, execs, positions) - providing a model independent and robust system.  Global coverage of equities and managed futures over 60+ markets.  Off-the-shelf alternatives struggled with the trade volume which peaked around 25,000 orders daily.  Not to mention my version being better value!  @Ashenden

Adapting systems to handle new models, markets, mandates, and everything else.  Supporting trade flow, operations, and infrastructure.  Front Office @MW, Ashenden

Model interfaces and services (prices, positions, execution, recs).  Eurohedge award.  @MW | Affinium

The broker has analytics but probably nicer to build your own and cross check.  Also you want to do pre and post-trade analysis at strategy level which your broker can't do.  Hat off to LehmanLive.  @Ashenden

An efficient way to scale.  @MW

You start off with one strategy trading into one fund ... and before you know it there are 30 strategies allocating into 20 funds.  Strategies and their dependent systems need scheduling ... unless you are already event-driven everything ;).  @MW @Ashenden

Models require adjusted historical market data in both sim and production modes.  These prices and volumes have undergone a multitude of adjustment factors owing to historical dividends and splits.  Building the framework, keeping the schedule updated, building the dynamic adjusted price series.  And then handling the adjustments which aren't always available at the time the model wants to trade (adjustment today impacting history from yesterday, which hasn't yet been published on BB before market open)!  @MW @Ashenden