牛顿投资管理

2023年4月28日

贸易炉边星期五…牛顿投资管理的尼尔·邦德贸易证明尼尔·邦德交易员和股票市场专家牛顿投资管理,对算法的发展空间,利用α生成算法目前面临的机遇和障碍。你看到什么新发展的算法空间?算法程序桌子以来进化非常发达处理大量的订单,现在他们甚至可以零售客户。机构客户空间,香草策略在很大程度上成为商品化和算法套件提供商似乎专注于改善智能路由器和顺序调整non-vanilla算法,试图模仿交易员的行为更复杂的订单——例如,扩展在关闭或切换策略在满足特定的条件。我也看到更多的工作做为了避免逆向选择,表明也许并不是所有的块流动性良好的流动性。最后,我认为更多的藻类提供商开始使用的供应商系统,他们觉得可以改善填充率,同时尽可能减少价格的影响。在藻类轮子可以用来更好的利用α创造机会吗?藻类轮子最初开发战略选择,但是吸收很快改变了(可能是受监管的担忧)作为工具来使用它们免除交易员的代理选择,避免代理偏好/交易员的偏见。我很高兴看到这一趋势正在向战略选择。他们是一个有用的工具来处理大容量/低副词%订单允许交易员关注订单可以增加更多的价值。 In order to generate alpha, maybe some portfolio manager profiling would be necessary but a large enough data set to analyse would be required. What lessons can other asset classes take from equity algo wheels? Other asset classes are embracing algos, but their market structures are so different from equities that I don’t think they will be using algo wheels to the same extent as in the equity world. Has algo usage reached its maximum capacity or do you expect it to account for more volumes in the years to come? Algos allow traders to handle vastly higher volumes than without and as such, I think they will continue to grow. I see traders using them more frequently for complex trades that would traditionally have gone to a high touch desk. For example, you no longer need a sales trader to access central risk book liquidity. They are constantly evolving and improving, which is a factor that you need to bear in mind when going down the customisation route – will your customised algo get upgraded at the same time as the non-customised algos? Also, all users, current and future, really need to know and understand what customisations have been made. I am seeing algo provider lists shrinking and stronger partnerships between users and providers being built. What are some of the hurdles that algos face? We are seeing growing retail participation in the markets, and it is very different from institutional flow with regards to the fact that it leads to higher volatility and price impact, together with relatively fast reversion and it is also largely inaccessible flow. It would be great if algo engines could factor this in and act accordingly, but it would be very useful to be alerted when retail participation is particularly high. Recently we had a bank crisis, and trading styles often change during such events, often toward more risk averse strategies like VWAP. When this happens, liquidity seeking strategies may be less likely to find blocks, so the strategies need to be able to adapt to a fast changing liquidity landscape.