Post

FICC003: Commodities - 能源、金属、农产品、期货曲线、库存与供需

FICC003: Commodities - 能源、金属、农产品、期货曲线、库存与供需

这是 PENGYI_FICC_MAPFICC003

按编号我们先做:

1
2
3
4
FICC000 -> FICC 总地图
FICC001 -> Fixed Income / Rates / Credit
FICC002 -> Currencies / FX
FICC003 -> Commodities

到这里,FICC 三大主线已经完整:

1
2
3
4
5
6
7
8
Fixed Income:
  money over time

Currencies / FX:
  money across countries

Commodities:
  physical supply-demand + futures curve + macro linkage

这一篇把第三块打透:

1
Commodities = 商品

公开边界:

1
2
This is educational material and research infrastructure thinking.
It is not investment advice, trading advice, or an actionable alpha note.

中文:

1
2
这是公开学习笔记,不是投资建议。
不包含内部数据、客户信息、未脱敏策略、实盘观点或可交易 alpha。

一句话总览

Commodities 的核心是:

1
实物供需、库存、运输、季节性、地缘政治、金融条件和期货曲线如何共同决定商品价格。

和股票、债券、外汇相比,商品最特殊的地方是:

1
它背后有真实物理约束。

例如:

1
2
3
4
5
原油要开采、运输、储存、炼化。
天然气要管道、LNG、储气库和天气需求。
铜要矿山、冶炼、库存、工业需求。
粮食要种植面积、天气、收成、库存消费比。
黄金要实际利率、美元、央行购金和避险需求。

所以商品不是简单的“价格时间序列”。

它是:

1
physical market + futures market + macro market + geopolitical market.

为什么 Commodities 属于 FICC

FICC 是:

1
2
3
Fixed Income
Currencies
Commodities

Commodities 属于 FICC,是因为它和宏观、利率、外汇、通胀、地缘政治、企业套保和机构风险管理高度相连。

商品影响:

1
2
3
4
5
6
7
8
9
inflation
trade balance
terms of trade
commodity exporter FX
producer margins
consumer cost
central bank reaction function
credit risk
portfolio allocation

例如:

1
2
3
4
5
6
oil price up
  -> headline inflation pressure up
  -> central bank may become more hawkish
  -> rates repricing
  -> USD / EM FX pressure
  -> credit conditions may change

或者:

1
2
3
4
5
6
copper demand weakens
  -> industrial cycle concern
  -> China/global growth concern
  -> risk sentiment weakens
  -> credit spreads may widen
  -> commodity exporter FX may weaken

商品是宏观周期的高频传感器。

商品三大类

Commodities 可以先分三大类:

1
2
3
Energy
Metals
Agriculture

进一步拆:

大类子类代表品种核心驱动
Energy能源crude oil, natural gas, gasoline, diesel, coal, power供需、库存、地缘、OPEC、天气、炼厂、运输
Metals金属gold, silver, copper, aluminum, nickel, iron ore实际利率、美元、工业周期、矿山供给、库存
Agriculture农产品wheat, corn, soybeans, cotton, sugar, coffee天气、种植面积、单产、库存消费比、出口、政策

还有一些交叉品类:

1
2
3
4
5
6
livestock
carbon / emissions
power
freight
lithium / battery metals
rare earths

但第一阶段先抓住:

1
2
3
4
5
oil
natural gas
gold
copper
wheat / corn / soybeans

这些足够建立商品框架。

商品和金融资产的区别

股票看:

1
2
3
4
5
earnings
growth
valuation
cash flow
discount rate

债券看:

1
2
3
4
5
cash flow
yield
duration
credit spread
default risk

外汇看:

1
2
3
4
relative rates
growth differential
capital flows
policy divergence

商品看:

1
2
3
4
5
6
7
8
9
physical supply-demand
inventory
storage
transport
weather
geopolitics
futures curve
producer / consumer hedging
macro overlay

商品难在:

1
它同时是实物资产、金融合约、宏观变量和地缘资产。

这也是为什么商品研究很适合 RAG + Graph。

因为很多信息不是结构化价格数据,而是:

1
2
3
4
5
6
7
8
9
OPEC statement
EIA inventory report
USDA WASDE report
mine disruption news
weather forecast
shipping bottleneck
sanction / conflict / export ban
refinery outage
policy announcement

Energy

Energy 是商品里最宏观、最地缘、最通胀相关的一块。

代表品种:

1
2
3
4
5
6
7
8
9
crude oil
natural gas
gasoline
diesel
heating oil
jet fuel
coal
power
LNG

能源研究最常见问题:

1
2
3
4
5
6
7
8
9
全球原油供需是否平衡?
OPEC+ 政策如何影响供给?
美国页岩油产量如何变化?
库存是在 draw 还是 build?
炼厂开工率如何?
成品油需求如何?
天然气天气需求如何?
LNG 贸易流如何变化?
地缘风险是否影响供应?

Crude Oil

原油是最核心的 energy commodity。

常见基准:

1
2
3
WTI
Brent
Dubai / Oman

研究原油要看:

1
2
3
4
5
6
7
8
9
10
11
12
global supply
global demand
OPEC+ production policy
U.S. shale production
inventories
refinery runs
exports / imports
spare capacity
geopolitical risk
shipping / sanctions
futures curve
time spreads

原油价格不是只由“经济好坏”决定。

它是多因素:

1
2
3
4
5
6
7
8
supply shock
demand shock
inventory cycle
financial positioning
USD
real rates
risk appetite
geopolitical premium

Oil Inventory

库存是油市的核心变量。

如果库存持续下降:

1
2
3
4
market may be tighter
physical demand may be stronger than supply
near-term scarcity premium may rise
curve may move toward backwardation

如果库存持续上升:

1
2
3
4
market may be oversupplied
physical demand may be weaker than supply
storage pressure may increase
curve may move toward contango

但不能机械理解。

要看:

1
2
3
4
5
6
7
8
9
10
11
crude inventory
gasoline inventory
distillate inventory
Cushing inventory
SPR changes
refinery utilization
imports
exports
production
seasonal average
market expectation

EIA 的 Weekly Petroleum Status Report 是公开能源研究的重要来源。

它提供:

1
2
3
4
5
6
7
8
crude oil stocks
petroleum product stocks
gasoline stocks
distillate stocks
refinery utilization
imports / exports
production
prices

对 Research OS 来说,这种报告非常适合做:

1
2
3
4
OilInventoryCard
EnergyBrief
CurveImpactNote
ForecastLedger

Natural Gas

天然气和原油不同。

它更受:

1
2
3
4
5
6
7
8
weather
heating demand
cooling demand
storage
pipeline constraints
LNG exports/imports
regional infrastructure
power generation

影响。

天然气更“区域化”。

因为运输和储存约束更强。

常见研究问题:

1
2
3
4
5
winter heating demand 是否超预期?
summer cooling demand 是否推高发电用气?
storage injection / withdrawal 是否偏离季节性?
LNG export capacity 是否改变本地平衡?
pipeline outage 是否造成区域价格冲击?

天然气是典型的:

1
weather-sensitive commodity

所以需要:

1
2
3
4
5
seasonality
weather forecast
storage report
infrastructure map
regional basis

Refined Products

原油不是终端消费品。

它要炼化成:

1
2
3
4
5
gasoline
diesel
jet fuel
heating oil
petrochemical feedstock

所以研究 oil 还要看 refinery。

关键变量:

1
2
3
4
5
6
7
refinery utilization
crack spread
gasoline demand
distillate demand
jet fuel demand
maintenance season
product inventory

crack spread 可以理解为:

1
refined product price - crude oil input cost

它反映炼厂利润和成品油市场紧张程度。

Metals

Metals 可以分为:

1
2
3
4
Precious metals
Industrial metals
Ferrous metals
Battery / energy transition metals

代表:

1
2
3
4
5
6
7
8
9
10
11
Precious:
  gold, silver, platinum, palladium

Industrial:
  copper, aluminum, zinc, nickel, lead, tin

Ferrous:
  iron ore, steel

Battery:
  lithium, cobalt, nickel, graphite

金属研究的核心是:

1
2
3
4
5
6
7
8
9
real rates
USD
industrial cycle
China demand
mine supply
inventory
energy cost
geopolitics
green transition

Gold

Gold 很特殊。

它不是普通工业品。

它的驱动包括:

1
2
3
4
5
6
7
8
9
real yields
USD
inflation concern
safe-haven demand
central bank buying
geopolitical risk
jewelry demand
ETF flows
positioning

黄金没有现金流。

所以它经常和:

1
2
3
real rates
USD
risk sentiment

联系紧密。

简化直觉:

1
2
3
4
5
6
real rates up
  -> opportunity cost of holding gold may rise
  -> gold may face pressure

real rates down or geopolitical risk up
  -> gold may be supported

但黄金不能用单变量解释。

例如央行购金、地缘风险、美元信用、避险需求都可能改变关系。

Copper

Copper 常被称为宏观周期敏感品种。

因为铜广泛用于:

1
2
3
4
5
6
7
construction
power grids
manufacturing
electronics
EVs
renewables
infrastructure

研究铜要看:

1
2
3
4
5
6
7
8
9
10
11
China demand
global manufacturing cycle
PMI
mine supply
smelter capacity
TC/RC
exchange inventories
scrap supply
energy transition demand
USD
risk sentiment

铜是典型的:

1
growth-sensitive commodity

所以铜价常被市场用来观察:

1
2
3
global industrial cycle
China stimulus expectation
manufacturing recovery

但也要小心。

铜价既有宏观需求,也有供给扰动。

不能简单说:

1
copper up = growth good

需要拆:

1
2
3
4
demand-driven move
supply-driven move
inventory-driven move
positioning-driven move

Iron Ore and Steel

铁矿和钢铁更接近:

1
China property / infrastructure / steel production cycle

关键变量:

1
2
3
4
5
6
7
8
China steel output
property construction
infrastructure demand
port inventories
mine shipments
steel margins
policy stimulus
environmental production cuts

这类商品非常受区域需求和政策影响。

Agriculture

Agriculture 是商品里最受天气和季节影响的部分。

代表:

1
2
3
4
5
6
7
8
wheat
corn
soybeans
rice
cotton
sugar
coffee
cocoa

农产品研究核心变量:

1
2
3
4
5
6
7
8
9
10
11
12
planted area
yield
weather
crop condition
harvest progress
exports
ending stocks
stocks-to-use ratio
biofuel demand
trade policy
currency
fertilizer cost

农业商品非常季节性。

例如:

1
2
3
4
planting season
growing season
harvest season
export season

不同月份市场关注点不同。

WASDE

USDA 的 WASDE 是农业商品研究的重要公开报告。

WASDE 全称:

1
World Agricultural Supply and Demand Estimates

它每月发布,提供美国和全球主要农产品的年度供需预测。

覆盖:

1
2
3
4
5
6
7
8
9
10
wheat
rice
coarse grains
oilseeds
cotton
sugar
meat
poultry
eggs
milk

对 Research OS 来说,WASDE 可以转成:

1
2
3
4
5
CropSupplyDemandCard
StocksToUseCard
WASDEChangeLog
AgricultureBrief
ForecastLedger

典型问题:

1
2
3
4
5
本月 WASDE 调整了哪个作物的 ending stocks?
产量调整来自 yield 还是 area?
出口需求是否被上调?
库存消费比是否偏紧?
市场预期和报告结果差在哪里?

Weather

天气是 agriculture 和 natural gas 的核心变量。

农业看:

1
2
3
4
5
6
7
drought
flood
heat wave
frost
rainfall
soil moisture
El Nino / La Nina

天然气看:

1
2
3
4
heating degree days
cooling degree days
winter storm
summer heat

天气不是单纯新闻。

它会进入:

1
2
3
4
5
supply forecast
demand forecast
inventory path
price volatility
option implied volatility

Futures

商品市场很大一部分通过期货交易。

Futures contract 是:

1
standardized contract to buy or sell an asset at a future date under specified terms.

商品期货重要,因为:

1
2
3
4
5
producers hedge future selling price
consumers hedge future purchase cost
traders express supply-demand views
investors access commodity exposure
markets reveal forward curve

期货不是简单“赌涨跌”。

它是:

1
risk transfer + price discovery + hedging + financing/storage signal.

Spot, Forward, Futures

三个基础价格:

1
2
3
4
5
6
7
8
Spot:
  当前现货价格。

Forward:
  双方约定未来交割价格,通常 OTC。

Futures:
  标准化交易所合约,每日盯市,有保证金。

商品研究经常看:

1
2
3
4
5
spot price
front-month futures
next-month futures
futures curve
calendar spreads

Futures Curve

期货曲线是商品研究的核心图像。

它描述:

1
不同到期月份期货合约的价格结构。

例如:

1
2
3
4
5
Jan contract
Feb contract
Mar contract
...
Dec contract

看曲线,不只是看绝对价格。

还要看:

1
2
3
4
5
6
near vs far
front spread
calendar spread
curve slope
curve shape
roll yield

Contango

Contango 通常指:

1
远月期货价格高于近月 / 现货价格。

简化图:

1
near price < far price

可能反映:

1
2
3
4
5
storage cost
financing cost
insurance cost
ample inventory
future supply-demand expectation

在 contango 中,如果持有多头并不断 roll 到更远月,可能面对负 roll yield。

但具体要看:

1
2
3
4
5
curve shape
roll mechanics
spot move
contract selection
transaction cost

不能机械说 contango 一定不好。

Backwardation

Backwardation 通常指:

1
近月期货价格高于远月。

简化图:

1
near price > far price

可能反映:

1
2
3
4
5
near-term physical tightness
low inventory
high convenience yield
strong immediate demand
supply disruption

Backwardation 常被视为现货紧张信号之一。

但也要谨慎:

1
不同商品、不同市场状态、不同合约设计下含义不同。

Cost of Carry

商品期货曲线和持有成本相关。

持有商品可能需要:

1
2
3
4
5
storage cost
insurance
financing
transportation
quality maintenance

这就是 cost of carry。

简化理解:

1
如果持有实物很贵,远期价格可能需要补偿这些成本。

但实际还要考虑:

1
2
3
4
convenience yield
inventory scarcity
physical demand
market expectations

Convenience Yield

Convenience yield 可以理解为:

1
持有实物商品带来的非现金便利收益。

例如:

1
2
3
炼厂持有原油,可以避免供应中断。
电厂持有燃料,可以保障发电。
生产商持有库存,可以稳定生产。

当现货紧张时,convenience yield 可能很高。

这会推动:

1
2
near-term prices higher
curve backwardation

Inventory

库存是商品研究的中心变量。

库存连接:

1
2
3
4
5
6
supply
demand
storage
curve
volatility
physical tightness

库存高:

1
2
3
市场有缓冲
供需冲击影响可能较小
contango 可能更常见

库存低:

1
2
3
市场缺少缓冲
小冲击也可能导致价格大幅波动
backwardation / volatility 可能上升

但库存要看:

1
2
3
4
5
6
7
absolute level
seasonal norm
days of demand cover
location
quality
accessibility
commercial vs strategic inventory

不是所有库存都一样。

Calendar Spread

Calendar spread 是商品期货研究常见对象。

例如:

1
front month price - second month price

或:

1
Dec contract - Mar contract

Spread 反映:

1
2
3
4
5
near-term tightness
storage economics
seasonality
roll pressure
physical market state

很多商品研究员比起 outright price,更重视 spreads。

因为 spreads 更直接反映:

1
physical supply-demand balance

Roll Yield

如果投资者通过期货持有商品 exposure,到期前需要换仓。

这叫:

1
roll

Roll yield 来自:

1
合约换仓过程中的曲线结构影响。

在 contango 中:

1
2
卖近月低价,买远月高价
roll yield 可能为负

在 backwardation 中:

1
2
卖近月高价,买远月低价
roll yield 可能为正

这对 commodity index 和 ETF 非常重要。

但实际表现仍取决于:

1
2
3
4
5
spot move
curve move
contract selection
roll schedule
cost

Commodity Basis

Basis 是:

1
local cash price - futures price

Basis 反映:

1
2
3
4
5
6
location
transportation
quality
storage
local supply-demand
delivery constraint

商品是实物市场,所以 location 很重要。

例如:

1
2
3
同样是天然气,不同区域价格可能差很多。
同样是原油,不同质量和交割地点会有差价。
同样是农产品,港口、产区、运输瓶颈都会影响 basis。

Supply-Demand Balance

商品研究最终经常落到供需平衡表。

基础结构:

1
2
3
4
5
6
Beginning inventory
+ Production
+ Imports
- Domestic consumption
- Exports
= Ending inventory

或者:

1
2
3
4
5
Supply:
  production + imports + beginning stocks

Demand:
  consumption + exports + ending stocks

关键不是公式复杂。

关键是判断:

1
2
3
4
哪个变量在变?
变化是否超预期?
市场是否已经 priced in?
库存路径是否偏紧?

Seasonality

商品有强季节性。

能源:

1
2
3
4
summer driving season
winter heating demand
hurricane season
refinery maintenance

天然气:

1
2
3
4
winter withdrawals
summer injections
heating degree days
cooling degree days

农产品:

1
2
3
4
5
planting
growing
pollination
harvest
export season

金属:

1
2
3
industrial production cycle
construction season
Chinese New Year effects

研究商品,不能忽略季节性。

否则会把正常季节变化误读成结构变化。

Geopolitics

商品和地缘政治高度相关。

典型事件:

1
2
3
4
5
6
7
8
9
10
war
sanctions
shipping disruption
export ban
pipeline sabotage
OPEC policy
trade restrictions
tariffs
strategic reserve release
port strike

地缘事件影响:

1
2
3
4
5
6
supply availability
transport routes
insurance cost
risk premium
inventory hoarding
price volatility

能源最明显,但金属和农产品也会受到影响。

USD and Rates

很多商品以美元计价。

因此美元和利率很重要。

1
2
3
4
5
6
USD stronger
  -> commodities may become more expensive for non-USD buyers
  -> commodity prices may face pressure

real rates higher
  -> opportunity cost for non-yielding assets like gold may rise

但不能机械。

如果商品出现强供给冲击,美元和利率影响可能被压过。

所以要区分:

1
2
3
4
macro-financial driver
physical supply-demand driver
geopolitical driver
positioning driver

Commodities and Inflation

商品是通胀的重要来源。

尤其:

1
2
3
energy
food
industrial inputs

商品价格上升可能推高:

1
2
3
4
5
6
headline CPI
PPI
transportation cost
input cost
household energy bills
food prices

但商品到通胀的传导要看:

1
2
3
4
5
6
7
weight in index
pass-through
substitution
tax / subsidy
currency
corporate margin absorption
policy response

这连接回 Fixed Income:

1
2
3
4
commodity shock
  -> inflation expectation
  -> central bank path
  -> yield curve

Commodity-Linked FX

商品和货币也强相关。

部分货币被称为 commodity-linked currencies:

1
2
3
4
5
6
7
AUD
CAD
NZD
NOK
BRL
CLP
ZAR

例如:

1
2
3
4
5
oil and CAD / NOK
iron ore and AUD
copper and CLP
agriculture and BRL
gold/platinum and ZAR

但这种关系也不是固定。

还要看:

1
2
3
4
5
6
rates differential
risk sentiment
domestic policy
capital flows
terms of trade
external balance

这就是为什么 FICC 三块不能割裂。

Commodity Research Workflow

一个 commodity research workflow 可以这样做:

1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
1. Define commodity
   oil / gas / copper / gold / wheat / corn / soybeans

2. Identify market structure
   spot, futures, physical market, key benchmarks

3. Supply side
   production, outages, policy, capacity, imports

4. Demand side
   consumption, industrial cycle, weather, exports

5. Inventory
   level, seasonal norm, location, draw/build

6. Futures curve
   contango / backwardation, calendar spreads, roll yield

7. Macro overlay
   USD, rates, growth, inflation, risk sentiment

8. Event risk
   OPEC, EIA, WASDE, weather, geopolitics, policy

9. Scenario
   base / bull / bear

10. What to monitor
   3-5 key indicators

这很适合做 Research OS。

Commodity Daily Brief Template

可以设计一个公开安全的商品日报模板:

1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
Date:

Commodity:
  oil / gas / gold / copper / soybeans

Market Move:
  spot:
  front-month futures:
  key spread:
  curve shape:

Supply Update:
  production / outage / OPEC / mine / crop / exports

Demand Update:
  refinery / power / industrial / China / weather / biofuel

Inventory:
  latest level:
  vs seasonal norm:
  draw/build:

Macro Overlay:
  USD:
  rates:
  risk sentiment:
  inflation:

Event Risk:
  EIA / WASDE / OPEC / weather / geopolitics

Interpretation:
  what changed?
  what matters?
  what to monitor?

Public-safe conclusion:
  no trading advice

这能成为:

1
FICC Daily Brief Generator

的一部分。

Energy Brief Template

能源专用:

1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
Crude:
  WTI:
  Brent:
  Brent-WTI:

Curve:
  front spread:
  6-month spread:
  contango / backwardation:

Inventory:
  crude:
  gasoline:
  distillate:
  Cushing:

Supply:
  OPEC:
  U.S. production:
  exports/imports:
  outages:

Demand:
  refinery utilization:
  gasoline demand:
  distillate demand:
  jet fuel:

Risk:
  geopolitics:
  sanctions:
  weather:
  policy:

Metals Brief Template

金属专用:

1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
Metal:
  gold / copper / aluminum / iron ore

Price move:
  spot:
  futures:
  spread:

Macro:
  USD:
  real rates:
  China growth:
  PMI:

Supply:
  mine output:
  smelter:
  disruptions:
  energy cost:

Demand:
  construction:
  manufacturing:
  grid / EV / renewables:

Inventory:
  exchange stocks:
  bonded stocks:
  warehouse trend:

Risk:
  policy:
  trade:
  sanctions:

Agriculture Brief Template

农产品专用:

1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
Crop:
  corn / wheat / soybeans / cotton

Report:
  WASDE / crop progress / export sales

Supply:
  planted area:
  yield:
  production:
  weather:

Demand:
  feed:
  food:
  biofuel:
  exports:

Inventory:
  ending stocks:
  stocks-to-use:
  vs expectation:

Seasonality:
  planting / growing / harvest / export

Risk:
  drought:
  flood:
  policy:
  currency:

Commodities x RAG

商品 RAG 可以处理:

1
2
3
4
5
6
7
8
9
10
EIA petroleum reports
USDA WASDE reports
OPEC statements
mine production updates
weather reports
company production guidance
port / shipping news
government policy documents
commodity exchange notices
research PDFs

典型问题:

1
2
3
4
5
本周 EIA 原油库存变化主要来自 supply 还是 demand?
WASDE 本月对玉米 ending stocks 的调整来自哪里?
铜库存下降是需求强还是供应扰动?
黄金近期变化更像 real rates 还是 safe-haven demand?
天然气库存路径是否偏离季节性?

RAG 输出必须包含:

1
2
3
4
5
6
source
timestamp
evidence
uncertainty
what to monitor
human review required

不能让模型凭空生成商品观点。

Commodities x Graph

商品非常适合 graph。

Energy graph:

1
2
3
4
5
6
7
8
crude oil
  -> OPEC policy
  -> U.S. shale
  -> inventories
  -> refinery runs
  -> gasoline / diesel
  -> inflation
  -> rates

Copper graph:

1
2
3
4
5
6
7
8
9
copper
  -> mine supply
  -> smelter capacity
  -> China demand
  -> construction
  -> power grid
  -> EV / renewables
  -> inventories
  -> USD / risk sentiment

Agriculture graph:

1
2
3
4
5
6
7
8
9
corn
  -> planted area
  -> weather
  -> yield
  -> production
  -> feed demand
  -> ethanol demand
  -> exports
  -> ending stocks

Graph RAG 的价值:

1
把事件、地区、库存、供需、价格、宏观变量连起来。

Commodities x Quant

公开安全地说,商品 quant 可以研究:

1
2
3
4
5
6
7
8
9
10
11
term structure signals
inventory surprise
seasonality
carry / roll yield
momentum
volatility
curve shape
cross-commodity relationships
macro factor sensitivity
event studies
forecast evaluation

但不能公开:

1
2
3
4
5
6
7
具体参数
实盘信号
未脱敏回测
交易团队观点
客户 flow
内部库存/报价
可复制 alpha

公开 demo 可以做:

1
2
3
4
5
6
commodity curve visualizer
EIA inventory brief generator
WASDE change summarizer
contango/backwardation explainer
gold-real-yield educational dashboard
copper macro graph

这展示的是:

1
2
research engineering ability
not proprietary trading edge

Commodities x Forecast Ledger

商品观点很适合做 forecast ledger。

每条观点记录:

1
2
3
4
5
6
7
8
9
10
11
12
timestamp
commodity
event
base view
bull scenario
bear scenario
evidence
market baseline
what to monitor
review date
outcome
post-mortem

例如:

1
2
3
4
5
6
7
8
9
10
11
event:
  EIA inventory draw larger than expected

view:
  near-term physical tightness increased

monitor:
  refinery utilization, exports, Cushing stocks, front spread

review:
  one week later

这不是交易建议。

这是研究判断的可审计化。

Pengyi Commodities Research OS v0

我们可以设计:

1
Pengyi Commodities Research OS v0

模块:

1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
Data Layer:
  public futures prices, curves, inventory data, macro data

Document Layer:
  EIA, USDA, OPEC, exchange notices, public reports

RAG Layer:
  source-grounded commodity document retrieval

Graph Layer:
  commodity -> supply -> demand -> inventory -> macro -> risk graph

Curve Layer:
  contango / backwardation / spreads / roll

Inventory Layer:
  draw/build, seasonal norm, location

Event Layer:
  weather, policy, geopolitics, production disruptions

Forecast Ledger:
  timestamped views and reviews

Brief Layer:
  energy brief, metals brief, agriculture brief

Human Review:
  PM / analyst approval before any conclusion

Artifact Layer:
  website note, chart, memo, slide

这可以和前面项目连接:

1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
LightRAG:
  commodity report memory

RAG-Anything:
  PDF/table/chart ingestion

GraphAgent:
  supply-demand relationship graph

FutureShow:
  forecast ledger and outcome review

Vibe-Trading:
  research workflow and artifact generation

MGP:
  memory governance and audit

和 Fixed Income 的连接

Commodities 和 Fixed Income 的连接主要通过:

1
2
3
4
5
inflation
real rates
central bank policy
growth expectation
risk sentiment

例如:

1
2
3
4
5
oil up
  -> inflation pressure
  -> rates repricing
  -> curve changes
  -> credit conditions
1
2
3
gold up
  -> real rates down / safe-haven demand up / USD down
  -> rates and FX context needed
1
2
3
4
copper down
  -> growth concern
  -> long-end rates may fall
  -> credit spreads may widen

所以不懂 FI,很难完整理解 commodities 的宏观含义。

和 FX 的连接

Commodities 和 FX 的连接主要通过:

1
2
3
4
5
6
terms of trade
export revenue
import cost
inflation
rate differential
risk sentiment

例子:

1
2
3
4
5
6
7
8
9
10
11
12
13
14
oil exporters:
  oil price up may support external balance

commodity importers:
  energy price up may worsen trade balance

AUD:
  linked to iron ore, China demand, risk sentiment

CAD / NOK:
  linked to oil, rates, global risk

CLP:
  linked to copper

这也解释了为什么 FICC002 的 FX 和这一篇必须连起来看。

商品和外汇是强连接。

面试可用表达

如果被问:

1
你怎么理解 Commodities?

可以回答:

1
2
3
4
5
6
7
8
9
10
11
我理解 Commodities 是 FICC 中最具有实物约束的一块。
它不只是价格时间序列,而是 physical market, futures market, macro market and geopolitical market 的结合。

我会先分成 Energy, Metals 和 Agriculture。
Energy 关注原油、天然气、成品油、库存、OPEC、炼厂、地缘和天气。
Metals 关注黄金、铜、铝、铁矿等,其中黄金更受 real rates、USD 和避险需求影响,铜更受工业周期、中国需求、矿山供给和库存影响。
Agriculture 关注天气、种植面积、单产、库存消费比、出口和 WASDE 等供需报告。

商品研究里我会重点看 futures curve、contango/backwardation、inventory、calendar spread、roll yield、supply-demand balance 和 seasonality。

我现在更感兴趣的是把商品研究 workflow 和 AI Research OS 结合起来,例如 EIA/WASDE RAG、commodity knowledge graph、curve monitor、inventory brief generator、forecast ledger 和 human-reviewed commodity research assistant。

这段可以用于:

1
2
3
4
5
6
FICC research support
commodities research
AI for finance
quant research interview
bank internal rotation
RA / PhD narrative

常见误区

误区一:

1
商品只看供需。

更准确:

1
商品看供需、库存、曲线、地缘、天气、仓储、运输、金融条件和持仓。

误区二:

1
库存下降价格一定涨。

更准确:

1
要看库存下降是否超预期、是否季节性、发生在哪里、质量如何、曲线是否确认 tightness。

误区三:

1
Contango 一定 bearish,backwardation 一定 bullish。

更准确:

1
它们反映 futures curve structure,常与库存、持有成本和现货紧张程度相关,但不能脱离具体商品和市场状态解释。

误区四:

1
黄金只看通胀。

更准确:

1
黄金要看 real rates、USD、央行购金、避险需求、ETF flows、地缘风险和市场结构。

误区五:

1
铜涨就一定说明经济好。

更准确:

1
铜价可能由需求、供给、库存、政策预期、美元、风险偏好和持仓共同驱动。

学习顺序

Commodities 初学顺序:

1
2
3
4
5
6
7
8
9
10
1. Spot / futures / forward
2. Futures curve
3. Contango / backwardation
4. Inventory
5. Supply-demand balance
6. Energy: crude oil / natural gas
7. Metals: gold / copper
8. Agriculture: WASDE / weather / stocks-to-use
9. Seasonality and event risk
10. Cross-asset link: rates / FX / inflation / risk sentiment

不要一开始就做复杂策略。

先把:

1
2
3
4
5
physical market
inventory
curve
supply-demand
macro overlay

这些地基打牢。

下一篇

现在 FICC 三大基础块已经齐了:

1
2
3
FICC001 -> Fixed Income / Rates / Credit
FICC002 -> Currencies / FX
FICC003 -> Commodities

下一篇最自然进入产品化和工程化:

1
FICC004 -> Daily FICC Brief Generator

也就是把 FI、FX、Commodities 三条线合成一个每日研究工作流:

1
2
3
4
5
6
7
8
macro calendar
rates / curve monitor
FX pair monitor
commodity inventory and curve monitor
news and event extraction
cross-asset causal chain
forecast ledger
human review

当前结论

Commodities 可以压成:

1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
Energy:
  oil, gas, refined products, inventory, OPEC, weather, refinery

Metals:
  gold, copper, industrial cycle, real rates, USD, China demand, mine supply

Agriculture:
  crops, weather, WASDE, ending stocks, stocks-to-use, exports

Market structure:
  spot, futures, forwards, curve, contango, backwardation, spreads, roll yield

Physical structure:
  supply, demand, inventory, storage, transportation, seasonality

Macro link:
  inflation, rates, FX, risk sentiment, geopolitics

Research OS:
  RAG + graph + curve monitor + inventory brief + forecast ledger + human review

这就是 FICC003 的核心。

商品不是简单的“价格涨跌”。

它是:

1
physical constraints + financial contracts + macro forces + geopolitical events.

这正好适合我们用 AI Research OS 去结构化、检索、追踪和复盘。

References

This post is licensed under CC BY 4.0 by the author.